Necessity 20 Years On: The Limits of Article 25

Abstract

Investment tribunals have fleshed out the requirements of the defence of necessity in Article 25 of the Articles on State Responsibility. This article considers the impact of necessity in international investment law and in general international law. The decided cases of the early twenty-first century have revealed three difficulties in Article 25 and in the defence of necessity. First, rather than assessing necessity from the vantage point of the State invoking necessity, they assess it ex post with hindsight bias. Second, the standard of proof required to demonstrate the existence of a grave and imminent peril is infeasible for many situations of risk and uncertainty. Third, the ‘only way’ criterion is not only unduly strict and virtually impossible to prove, but is wholly unrealistic for macro-level crises, such as pandemics or financial crises. To account for these difficulties, tribunals should avoid hindsight bias, especially with respect to ‘grave and imminent peril’ and the ‘only way’ criterion. The defence of necessity is a blunt instrument and has not afforded States a significant ‘safety valve’. As a result, States have begun to reconsider the substantive standards in investment treaties and to include internal exceptions in their investment treaties.

I. INTRODUCTION

Investment tribunals have been instrumental in fleshing out the requirements of the defence of necessity in Article 25 of the Articles on State Responsibility (ARS).

This provision states as follows:

• Necessity may not be invoked by a State as a ground for precluding the wrongfulness of an act not in conformity with an international obligation of that State unless the act:

  • is the only way for the State to safeguard an essential interest against a grave and imminent peril; and

  • does not seriously impair an essential interest of the State or States towards which the obligation exists, or of the international community as a whole.

 

• In any case, necessity may not be invoked by a State as a ground for precluding wrongfulness if:

  • the international obligation in question excludes the possibility of invoking necessity; or

  • the State has contributed to the situation of necessity.

Now, twenty years after the International Law Commission (ILC) adopted the ARS, it is an opportune moment to consider the role and limitations of necessity in investor-State arbitrations to date and the impact of this defence on international investment law as well as general international law. Necessity is a defence of last resort that can supplement the existing flexibility built into treatment standards in investment treaties. In practice, moreover, the defence has also served to underscore that existing primary rules are insufficiently flexible.²

The requirements of necessity are abstract and general. The ILC articulated them at the end of the twentieth century on the back of few instances of practice. This practice mostly endorsed the principle that superior interests took priority over inferior interests in situations of grave and imminent danger to essential interests. Commenting on the ILC’s task in codifying the law of State responsibility, James Crawford noted that elaborating general rules based on ‘case law, practice and often unarticulated instantiations of general legal ideas’ sometimes involved ‘inventing them’.³ Indeed, draft Article 33, adopted in 1980, was the first institutional formulation of the defence of necessity in international law.⁴ The engagement of investment tribunals with this plea has provided some needed clarification on the meaning of necessity’s, it’s general and abstract requirements. The case law has thus provided a testing ground for the ILC’s formulation of this defence and showed areas where the defence falls short, particularly, as we argue, when applied to macro-crises, which affect various aspects of the functioning of States and which require multiple measures.

Section II considers how investment tribunals have applied the defence of necessity over the last 20 years. Section III draws broader lessons from this case law for international investment law and analyses how the shortcomings of the defence of necessity have affected the drafting of investment treaties. A growing recognition by States of the need for more regulatory space under investment treaties is expressed in a shift to internal exceptions in addition to, or in lieu of, necessity (and other defences). This shift, we argue, demonstrates the lawmaking function of necessity, in its ordinary meaning (and not as a general defence). Section IV concludes and looks to the future of the plea of necessity in general international law.

II. THE CONTRIBUTION OF INVESTMENT TRIBUNALS TO CLARIFYING THE REQUIREMENTS OF NECESSITY

This section reviews the state of the art on the plea of necessity as it has emerged since 2001. Most of the considered engagement with the defence of necessity over the last two decades has been by investment tribunals.⁵ For the most part, the published investment awards involve necessity defences invoked in the context of the Argentine financial crisis in 2001, though States have also invoked necessity in the context of war,⁶ revolutions,⁷ national security crises⁸ and public order and security.⁹

There are three overarching themes that inform our analysis. The first is the rationale of the defence of necessity. Necessity in Article 25 ARS reflects a lesser-evil logic: in a situation of danger or threat where two interests are in conflict, such that harm to one of the two interests appears probable, necessity exonerates the invoking State for avoiding the greater harm to its interest if it is the superior one in the circumstances. By safeguarding the superior interest in the circumstances, the State minimizes overall harm. This consequentialist premise casts the plea as the ‘paradigmatic justification’.¹⁰ The lesser evil is one of the main theories of justification, meaning that the plea of necessity renders the action which harms the lesser interest—all things considered—permissible and, thereby, lawful. Some investment tribunals consider necessity to be an excuse,¹¹ and while this would be a welcome development both normatively and practically,¹² it is one that States have yet to endorse. In investment arbitration, Argentina,¹³ Bolivia,¹⁴ Egypt¹⁵ and Zimbabwe¹⁶ have argued necessity as a justification.¹⁷ We assume in this article that necessity is a justification, premised on a lesser-evil rationale.¹⁸

Second, the plea of necessity is about avoiding future harm, and hence requires States to predict the future. For some harms, the probability of the harm occurring is known or at least readily estimable. For others, this probability is uncertain. The distinction between risk and uncertainty in a setting of incomplete information¹⁹ is central for our analysis—particularly for the criteria of ‘grave and imminent harm’ and ‘only way’. Risk is ‘measurable uncertainty’ or quantifiable uncertainty.²⁰ With risk, we know the probability of an event and can predict expected harm.²¹ States and other actors can manage risks,²² but may err in assessing such risks due to heuristics and biases.²³ By contrast, States cannot quantify uncertainty because the probability distribution is unknown/unknowable.²⁴

This distinction between risk and uncertainty goes back to economists Keynes and Knight,²⁵ and is also known as Knightian uncertainty.²⁶ In 1921, Knight underscored that ‘uncertainty must be taken in a sense radically distinct from the familiar notion of risk … it is “[t]rue” uncertainty’.²⁷ Nelson and Katzenstein explain the distinction as follows:

In risky environments, sorting events into different classes poses no special challenge for sophisticated decision makers. We cannot be sure what tomorrow will bring, but we can rest assured that unforeseen events will be drawn from known probability distributions ‘with fixed mean and variance’. In the world of risk the assumption that agents follow consistent, rational, instrumental decision rules is plausible. But that assumption becomes untenable when parameters are too unstable to quantify the prospects for events that may or may not happen in the future. The past is not a prologue. Realms of uncertainty are subject to dramatic transformations in the underlying economic structure that permanently shift the mean of the distribution. In this new environment there is no basis for agents to settle on what the ‘objective’ probability distribution looks like.²⁸

As John Kay and Mervyn King note, ‘most of the problems we face in politics, business (including finance) and society’ involve [radical] uncertainty.²⁹ Decision makers thus often face (difficult) choices under uncertainty. They ‘know something, but not enough to enable [them] to act with confidence’.³⁰ For example, in a financial crisis decision makers do not know ex ante what package of macroeconomic and other measures will stabilize the economy. Similarly, when the WHO declared the SARS-CoV-2 virus a Public Health Emergency of International Concern (PHEIC) on 30 January 2020,³¹ decision makers knew little about the novel virus and operated under uncertainty as to what package of measures would work in response.³² Modelling of the novel pandemic was then still in its infancy and initially based on very limited data.³³ There was a wide range of possible futures. It was uncertain for example whether the virus would cause 10,000, five million or one billion deaths worldwide. Indeed, as David Spiegelhalter and Anthony Masters note, ‘epidemic models are full of uncertainty’, flowing from ‘their simplified structure, their assumptions and data inputs and the unpredictability of real life’.³⁴

Third, adjudicators assess whether the State was justified in invoking necessity after the fact, with the benefit of hindsight. The ILC Commentary does not make this explicit, but it suggests at multiple points that adjudicators must assess the defence from the vantage point of the State at the time it adopted the relevant measures.³⁵ This is an objective determination: it does not refer to what the State knew or perceived (‘apprehended’), but rather what the State could reasonably know at that time. Tribunals trying to place themselves in the same situation as the invoking State in this assessment have knowledge that the State could not have had at the time. They must thus guard against hindsight bias.³⁶ Hindsight bias refers to the human tendency to attach an excessively high probability to an event simply because it occurred.³⁷ Hindsight bias can be very difficult to correct as it tends to persist even when adjudicators (and others) make considered efforts.³⁸

In our analysis below, we consider the following elements of necessity in turn: (i) grave and imminent peril; (ii) essential interest; (iii) the ‘only way’; (iv) non-contribution; (v) the temporal dimension of necessity (its beginning and end); and (vi) the duty of compensation if a State successfully invokes the plea. We do not address the question whether the specific substantive rule excludes reliance on the plea. This is a question mainly about the interpretation of the relevant substantive rule and not about the plea of necessity.

A. Grave and Imminent Peril

The plea of necessity arises in situations where there is a grave and imminent peril to an essential interest of the State. Despite its seemingly simple formulation, the concept of ‘grave and imminent harm’ hides complexities. There are two important aspects of this element: first, the meaning of the concept of ‘grave and imminent peril’ and, second, questions of proof as to the existence of this peril, given how the ICJ and the ILC have interpreted ‘imminent’. When harm is uncertain, there can be no grave and imminent peril. Each aspect raises different issues, so we address them separately below.

(i) The meaning of grave and imminent peril

The term peril refers to the ‘position of being imminently exposed to the chance of injury, loss, or destruction’.³⁹ The concept of peril can therefore be broken down into two component elements: risk and harm.⁴⁰ To meet this requirement, the State must demonstrate the existence of a risk of a grave and imminent harm to an essential interest.⁴¹ The risk can refer both to the occurrence of ‘triggering events’ which could cause grave and imminent harm to an essential interest (the adverse effect in the form of material damage), and to the grave and imminent harm itself.⁴² The triggering event can be the occurrence of a natural hazard (eg an earthquake or a pandemic) or a human act (eg armed conflict), and it can occur inside or outside the State’s boundaries.⁴³ Often, risks emanate from a combination of these factors.⁴⁴ The threatened harm to the essential interest must be grave, in a quantitative or qualitative sense. Furthermore, it must be imminent. The word ‘imminent’ does not require that the risk is about to materialize (immediacy in temporal terms);⁴⁵ rather, it has been interpreted as harm not yet having (completely) materialized at the time a State acts in a situation of necessity.⁴⁶ This is in line with the rationale of the defence that allows the State to take measures to prevent grave harm.

Three aspects of the condition of imminence require elucidation. First, at the time the State adopts measures of necessity it acts based either on a forecasted harm (which can be expressed in probabilities) or uncertain harm (which cannot), rather than actual harm.⁴⁷ Situations of risk and, especially, of uncertainty pose major difficulties for tribunals assessing necessity pleas, even if the harm is certain/has materialized at the time of adjudication and adjudicators benefit from hindsight.⁴⁸ Tribunals need to assess harm based on the information reasonably available to decision makers at the time of decision-making: they otherwise risk holding the State against a standard that is impossible to meet.⁴⁹

In situations where the harm remains uncertain at the time of adjudication, necessity is even more difficult, if not impossible, to invoke. As Caroline Foster explains, ‘precaution involves a preparedness by public authorities to intervene in advance in relation to potential, uncertain or hypothetical threats’.⁵⁰ Such scenarios are at the outer limit of necessity, and stray into the domain of the precautionary principle.⁵¹ The precautionary principle seeks to address scientific uncertainty about future harms.⁵² In such scenarios, neither decision makers nor tribunals can base their decisions on science because scientific research is insufficiently advanced. The precautionary principle calls for a ‘move away from the “primacy of scientific proof” and recognising the limits of science’.⁵³ The defence of necessity does not presently accommodate such uncertainty that persists at the point of adjudication.

It is not clear if there is a maximum duration between the identification of the risk of harm and the harm materializing for the plea of necessity to apply. The Commentary to Article 25 indicates that the harm must be ‘proximate’.⁵⁴ ‘Proximate’ can be understood in temporal or causal terms. The ILC seem to take a causal understanding of ‘proximate’: quoting approvingly from Gabčíkovo–Nagymaros, the ILC acknowledged the possibility of long-term harms which could be sufficiently established.⁵⁵ In Gabčíkovo–Nagymaros, the Court rejected Hungary’s invocation of ecological necessity in such a scenario not because the harm was far in the future, but because the harm was not sufficiently established.⁵⁶ However, where the time lag between the ascertainment of the risk and the materialization of the harm is great, States may struggle to convince tribunals that their actions meet another requirement of necessity: the ‘only way’. Indeed, our ability to assign a degree of probability to the occurrence of the harm tends to diminish the longer the forecast horizon is: the further the harm is in the future, the more difficult it will be to estimate the probability of its occurrence. In addition, when the threatened harm is well in the future, States will usually have time to identify adequate means, including cooperation, to prevent them. So, while the Commentary, on the basis of Gabčíkovo–Nagymaros, seems to support an understanding of ‘proximate’ as purely causal, it seems that some temporal proximity is required as well: the State must be in a position where it needs to act at that point in time to avert the risk of harm. Differently put, it is not necessary that there is a risk of immediate harm, but there must be a need for immediate action.

Second, the defence is clearly available when the harm is wholly in the future. The plea of necessity can be reactive (the triggering event has occurred or is unfolding, and the State acts to prevent the materialization, or reduce the impact, of its consequences) or prophylactic (the event has not occurred yet, and the State acts to prevent its occurrence or pre-emptively guard against its consequences). When the harm has commenced to materialize but is continuing,⁵⁷ the defence should be available to avoid further aggravation as this minimizes overall harm.⁵⁸ By contrast, if the harm is entirely in the past, the plea should no longer be available.

Third, temporal relations between the triggering event and the harm can vary. While the harm to the interest is in the future (wholly or partially), the triggering event can itself be in the past, be unfolding or be in the future. In the Torrey Canyon incident, damage to the leaking ship (a past event) threatened the UK’s coasts with future environmental harm; the UK thus took measures to prevent the harm from materializing.⁵⁸ In Gabčíkovo–Nagymaros the completion and operation of works in the Nagymaros sector (a future event) threatened to harm the people and environment of Hungary (a future harm).⁶⁰ By suspending work, Hungary sought to prevent the occurrence of the triggering event and, thereby, to neutralize the harm. Similarly, the arrival of the SARS-CoV-2 virus on the territory of many States and community transmission (an unfolding event) threatened the population (a present and future harm).

The investment case law has highlighted the difficulties of the temporal relations between risk and harm. What is required is that at the time of the act in necessity the risk is sufficiently established,⁶¹ such that action is needed to avert it: it is not necessary that the triggering event has occurred or that the harm has begun to materialize. The Argentine financial crisis cases are illustrative. Some tribunals held that the financial crisis, alone or in combination with the attendant political and social instability, were the unfolding triggering events which generated the threat of grave harm to the State’s interests.⁶² Other tribunals instead demanded a ‘total collapse’ (of the economy, of the State’s institutions) as the triggering event.

Conceptually, both approaches work: the ‘total collapse’, whether a past, unfolding or future event, can generate a threat of harm for the State’s interests. In practice, it makes no difference for whether necessity is available which event (the financial crisis or the total collapse) is the trigger. If either event is sufficiently established (on which more below) at the time the State is acting, and a causal connection can be established or forecasted between that event and the harm to the State’s essential interests, this element of the plea is met. In keeping with this approach, the LG&E Tribunal allowed the plea of necessity because the financial crisis, and attendant public disorders, could result in a ‘total collapse’ (future event) which would harm Argentina’s essential interests (future harm).⁶³ Not all tribunals appreciated this, and some denied the plea because the selected triggering event had yet to occur. Thus, the Tribunal in CMS rejected the plea because, although the crisis was severe, it had not yet resulted in ‘total economic collapse’.⁶⁴ This is an unnecessary strict, even illogical, reading of the defence of necessity. If the purpose of the defence is the protection of superior interests, then the State must be allowed to act to avert the triggering event as well as to avert future harm to its interests by that event.

(ii) Establishing grave and imminent peril

In Gabčíkovo–Nagymaros, the ICJ required that the invoking State ‘sufficiently establish’ that the peril was ‘certain and inevitable’ and not ‘merely apprehended or contingent’.⁶⁵ According to the Court, by the time Hungary abandoned the dam, it had not met this burden (nor shown ‘imminence’ of the harm).⁶⁶ This standard is unduly restrictive and is incompatible with the premise of the plea: to justify State action in the event of risk. Risks cannot be established with certainty: they refer to future events and harms and, at most, we can attach some probability to their occurrence. The standard should therefore be one of probability of the event or harm occurring,⁶⁷ rather than certainty or inevitability. Moreover, a standard of certainty or inevitability is one that implies an ex post assessment of this requirement from the vantage point of the adjudicator. Only hindsight tells us whether an event or harm was certain or inevitable.

The ILC allowed space for risk and uncertainty within necessity, and for the assessment to be made from the vantage point of the State (although still on an objective basis): ‘a measure of uncertainty about the future does not necessarily disqualify a State from invoking necessity, if the peril is clearly established on the basis of the evidence reasonably available at the time’.⁶⁸ The ILC’s standard of proof is thus stated in the negative: some measure of uncertainty does not exclude the defence. This is a better standard, though it still fails to indicate the degree of uncertainty beyond which the defence is excluded. The Commentary states that peril must not be merely ‘apprehended’ (subjectively perceived by the State), nor must it be merely ‘contingent’ (down to chance).⁶⁹ But is it enough for the State to prove the peril on the balance of probabilities?⁷⁰

B. Essential Interests

The lesser-evil logic of necessity requires weighing two conflicting interests: one of the two interests must be sacrificed. The plea of necessity ensures that the State can permissibly act to protect the interest which is deemed superior in the for doing so avoids a greater harm. This requires a comparative exercise between the interests at stake. The ILC does not define, nor list, interests that are essential. It simply states that whether an interest is essential for the purposes of the plea ‘depends on all the circumstances, and cannot be prejudged’.

The analysis of essential interests can usefully be divided into three steps: (i) the interest safeguarded; (ii) the interest impaired; and (iii) weighing these interests and the harms to these interests.

(i) Interest safeguarded

A State can act to protect ‘an essential interest’, which may include its own interests, those of its people and those of the international community as a whole.⁷¹ The Commentary contains little guidance on how to identify essential interests: it says only that depends on ‘all the circumstances, and cannot be prejudged’.⁷² The drafting of Article 25 indeed reveals that the Commission intended this to be a relative concept.⁷²

The defence of necessity is not limited to the survival of the State, unlike the historical practice on the right of self-preservation.⁷⁴ For example, the ICJ in Gabčíkovo–Nagymaros accepted that ecological interests could be essential for the purposes of the plea.⁷⁵ The ILC’s Commentary does not provide a list of essential interests: the examples referred to therein are intended to illustrate the generality of the defence and not to provide a list of which interests may count as essential.⁷⁶ James Crawford has explained that the etymology of the word ‘essential’ ‘suggests a connection to the “life” of the State’.⁷⁷ Since most interests of the State can plausibly be explained as related to its ‘life’, what really seems to matter is the extent of the harm threatened to the interest in question. An extremely grave harm to a financial interest, for example, can affect the life of the State, as it may bring the whole State apparatus to a halt.

Most investment tribunals have endorsed this view. They have qualified the following State interests as essential in the circumstances: the existence and independence of the State,⁷⁸ the maintenance of public order,⁷⁹ the well-being of the State’s population, access to public services, and the functioning of public institutions.⁸⁰ Tribunals have also reviewed State assertions as to the essential character of the interests relied on. In von Pezold, the Tribunal found that the interest relied on in that case was ‘ensur[ing] the survival of the incumbent Government and its President at a political level’, and this was neither a ‘threat to the State itself’,⁸¹ nor did it concern a State interest. The conclusion could have been better supported by reference to the circumstances: the collapse of a government could affect the life of the State by causing political and social instability, but that threat was not imminent. In Unión Fenosa v Egypt, the Tribunal rejected the Respondent’s view that the relevant interest that its measures protected was ‘the shortage of gas in Egypt’. The Tribunal held this was not an essential interest, as the harm to it had resulted from Egypt’s own policies.⁸² The reasoning is faulty: the origin of the harm, although relevant to the requirement of non-contribution, is immaterial to the essential interest.

(ii) Interest impaired

The second limb of the analysis involves the interest impaired. Article 25 requires that the act in necessity does not ‘seriously impair an essential interest of the State towards which the obligation is owed or of the international community of States as a whole’. It is reasonable to assume that the method for determining whether the interests impaired are ‘essential’ is the same as for the interest safeguarded. Namely, the essential character of the interest depends on ‘all the circumstances’. Here, too, it seems that what matters are the circumstances and the extent (the seriousness) of the harm to this interest.

Investment tribunals have dealt with this requirement with some hesitation. Tribunals have struggled to identify whose interests ought to count in this limb. They have identified two broad interests: the interest of the home State in its investment-treaty rights and/or the interests of the investor,⁸³ though without properly engaging with this question. Article 25 ARS speaks of the interests of the ‘State towards which the obligation exists’ or the international community as a whole.⁸⁴ And yet, Part One of ARS is applicable not just to inter-State obligations.⁸⁵ Whether the interests of non-State actors,⁸⁶ to which Article 25 does not specifically refer, can be considered in this calculus remains unresolved and, much like the case law on countermeasures,⁸⁷ the case law on necessity has so far failed to provide answers.

(iii) Weighing interests

The ILC Commentary states that the interest protected ‘must outweigh all other considerations, not merely from the point of view of the acting State but on a reasonable assessment of the competing interests, whether individual or collective’.⁸⁸ The defence of necessity allows a State to not perform ‘some other obligation of lesser weight or urgency’.⁸⁹ This means that once tribunals have identified the interests, they must weigh them against each other: only after this weighing can they determine which of the two interests ought to be safeguarded (and which harmed) in the effort to minimize the overall harm. The weighing must consider the respective gravity of the harms that each of the interests is either expected to or has in fact suffered. If the interests are of equal weight, or even if the interest impaired is qualitatively weightier, tribunals must resolve the weighing by reference to the respective harms expected or caused.

The Commentary to Article 25 does not clarify the vantage point from which the weighing must be performed and whether a tribunal may consider the actual harm caused to the interest impaired by the measures of necessity. The drafting history of this provision may provide an answer. The commentary to draft Article 33 (the predecessor to Article 25) indicated that ‘any conduct going beyond what is strictly necessary for this purpose will inevitably constitute an internationally wrongful act per se, even if the excuse of necessity is admissible as regards the remainder of the conduct’.⁹⁰ This statement suggests that adjudicators ought to take the vantage point of the invoking State: that is, when weighing the interests, Tribunals must take into account the forecasted harms to each of the interests.⁹¹ This vantage point avoids holding States accountable under a standard that was impossible for them to meet and avoids hindsight bias by the adjudicators. The invoking State could be held pro tanto liable for the harm caused in excess of its forecast, as indicated in the commentary to draft Article 33. However, where the actual harm to the impaired interest involves an essential interest and is ‘serious’, the scale could tip in the other direction and exclude the plea (for, ultimately, a greater harm has been caused). This solution involves hindsight assessment by the adjudicators but minimizes overall harm, as well as reducing the risk of abuse that is inherent in the defence.

It is easy to see why this aspect of the plea of necessity has drawn (both historically and presently) much criticism and concern.⁹² How can this weighing be performed, given that there is no absolute scale of interests or values in international law, save for interests protected by peremptory rules? By what benchmark can we determine that protecting one interest and sacrificing another would minimize harm? Does it matter how acutely a State perceives the risk to be, knowing that social and cultural factors play a role?⁹³ Are all harms commensurable? And how can probabilities be compared?⁹⁴ Critics have directly challenged the choice of lesser-evil as the blueprint for this defence.⁹⁵ This formulation is typical of domestic criminal law justificatory necessity.⁹⁶ Often, domestic necessity defences specify the interests in respect of which they can be invoked (eg life or limb), and against which they cannot be invoked (usually, life).⁹⁷ In these cases, legislators have performed the weighing ex ante. In all cases, regardless of ex ante weighing by legislators, courts review domestic claims of necessity by individuals ex post. Yet difficulties remain: in domestic law, too, problems of incommensurability can arise where the defence does not specify the relative weight of values in advance.⁹⁸ There is also a perceived ‘democracy problem’: should the judiciary decide, on a case-by-case basis, which interests ought to be regarded as superior in a given society, or should this be a task for legislators?⁹⁹

The impossibility of weighing interests and harms, critics say, makes the defence unworkable in international law,¹⁰⁰ leaving it open to the utmost subjectivism by States which are wont to consider their interest as more essential than others’ interests. Indeed, in international law, invoking States weigh the relevant interests on a case-by-case basis, and ex post oversight is sporadic. What is more, who gets to decide what interests should be considered as superior? Is this a task for adjudicators, or is there here too an equivalent of the ‘democracy problem’ mentioned earlier—perhaps in the form of a ‘legitimacy problem’? On the one hand, by affording some degree of deference to the weighing performed by the State invoking necessity, adjudicators respond to the legitimacy problem¹⁰¹ and avoid hindsight bias. On the other, too much deference may be incompatible with the basic principle of sovereign equality: can one State really decide on the relative importance of the interests of other States? The degree of deference afforded may depend on the legal context in which necessity is invoked, for example in cases where the interest harmed is evidently not essential as it is not, in Crawford’s words, ‘connected to the life of the State’, such as investment law.

Despite many of the above criticisms, investment tribunals have not had much difficulty with the weighing of interests. Their conclusions on this point are, on the whole, unobjectionable: could it be said that the well-being of a State’s population, or the continued functioning of public services, was not superior to the interests of investors and/or their home States? But their reasoning has not been satisfactory. In addition to lack of clarity as to whose interest must be counted in the second limb, the investor’s home State or the investor, there has been no real engagement with the weighing exercise and, on occasion, no acknowledgement or justification for why investment tribunals skipped the second limb of the analysis or the weighing operation altogether.¹⁰² This may have resulted from the context and tribunals’ intuition as to the relative ‘weight and urgency’¹⁰³ of the interests in question, and perhaps to a perceived, but unstated, desire to defer to the invoking State’s assessment on this matter. But the seeming ease with which investment tribunals arrived at these conclusions is unlikely to be replicated in other, and in particular inter-State, dispute settlement settings.

C. The ‘Only Way’

Article 25(1)(a) states that the act must be ‘the only way for the State to safeguard’ the relevant interest. The ILC’s Commentary explains that necessity is excluded if ‘there are other (otherwise lawful) means available, even if they may be more costly or less convenient’.¹⁰⁴ ‘Costly’ refers to financial cost: the State needs to prioritize an alternative, lawful measure that is more expensive over the prima facie unlawful measure.¹⁰⁵ Less convenient refers to measures with increased administrative or organizational burdens. The ‘only way’ is, without doubt, a strict requirement. To quote Crawford: ‘Here “only” means “only”’.¹⁰⁶

This element has caused the most difficulty in the investment case law and is usually the element on which the defence fails.¹⁰⁷ Most investment tribunals have nevertheless glossed over this element. In the cases on Argentina’s financial crisis, tribunals often limited themselves to noting that expert reports and comparative analyses with crisis management in other States showed that there were alternative measures to tackle the crisis and that, for this reason, this criterion was not met. Such a restrictive approach to this criterion renders the defence illusory.¹⁰⁸ By the same token, the case law has highlighted the deep complexities of this element and called into question its suitability as currently understood. Without doubting the need for a strict formulation of the defence in the name of avoiding its abuse, the ‘only way’ criterion expects States to know which measures will not work and to identify the ‘only one’ that works. The premise is that only ‘one way’ that works exists. Yet most crises will require a combination of measures, and States can only estimate their effectiveness at the time of decision-making. Whether a measure was the only way becomes clear only in hindsight.

There are, in our view, three relevant parameters in assessing the ‘only way’ requirement, which roughly track those that the Enron Committee identified.¹⁰⁹ These parameters are at times implicit, but often absent in the reasoning of tribunals: (i) the focus of analysis (a single measure or a package of measures); (ii) the relevant vantage point for the assessment (the problem of hindsight); and (iii) the availability of alternative measures.¹¹⁰ For ease of analysis, we consider these three interrelated parameters separately.

(i) A single measure or package of measures?

The first parameter is the focus of assessment. Investors might challenge a specific measure because the State has adopted only one measure and because that measure impairs investor rights under investment treaties. In von Pezold, for example, Zimbabwe adopted the Fast Track Land Reformation Programme and the question was whether that measure was the only way to stop occupation of farmland by settlers to stabilize the country.¹¹¹

Other times, however, the State may address a crisis with a package of measures, and investors might consider that only one measure or subset of all the measures impaired their entitlements under investment treaties. The Argentine financial crisis cases illustrate this scenario. In response to its financial crisis 20 years ago, Argentina adopted a package of measures from mid-2000 to mid-2002. Argentina abandoned the 1:1 currency peg between the peso and the US dollar in 2001 and forced conversion of dollar-denominated assets and liabilities into pesos, it tightened public spending, increased taxes, limited withdrawals from deposit taking institutions, introduced capital controls and restructured its sovereign debt.¹¹²

The approach of investment tribunals in the Argentine financial crisis cases has varied considerably. While some have focused on the discrete measure challenged,¹¹³ others have taken a broader outlook and considered the package of measures.¹¹⁴ The former approach is too narrow, and the latter, at least as interpreted in LG&E (ie the State had to do something, and it did something) is too broad. The approach of the Urbaser Tribunal charts an appropriate middle way:

The emergency measures and the state of necessity associated with them were events of nation-wide importance. Therefore, the question whether ‘other means’ were available has to be captured in both perspectives: the wide one, taking into account the needs of Argentina and its population nation-wide, and the narrower one of the situation of investors engaged in performing contracts protected by the international obligations arising out of one of the many BITs.¹¹⁵

This approach underscores that the specific measure challenged must be understood in the context (and as an integral part of) the State’s response to the crisis. Each measure individually is unlikely to be the ‘only way’—indeed, the very fact that the State’s response is composed of a package of measures suggests that multiple measures are available (and perhaps necessary or reasonable). Such a narrow focus misses the point: macro crises require multipronged responses.¹¹⁶ To assess a single measure of this package alone is artificial. Framing the question as focused on a single measure already presupposes the answer. As the Argentine investment tribunals have shown, when the question is posed in this way the answer, invariably, is ‘no, it is not the only way’.¹¹⁷

(ii) The problem of hindsight

Tribunals often assess the plea of necessity years after the onset of the crisis and the adoption of the measures in question. Should tribunals take advantage of this hindsight or should they judge the measures from the standpoint of the State at the time of adopting the measure? For the most part, the Argentine financial crisis tribunals seem to have approached this assessment with the benefit of some or all of the information available after the fact. This has led tribunals to find that because the measures were ineffective, they were not the only way.¹¹⁸ Conversely, even if there had been only one way ex ante, the success of the measures has led adjudicators to conclude that ex post they were not necessary at all for no crisis materialized in the end (which concerns not just the only way requirement, but also whether there was a grave peril to begin with).¹¹⁹ These conclusions are unreasonable.

Other, though fewer, tribunals have assessed this question from the standpoint of the State at the time of adopting the measure.¹²⁰ As the Continental tribunal put the challenge of hindsight, ‘a time of grave crisis is not the time for nice judgements, particularly when examined by others with the disadvantage [or better with the benefit] of hindsight’.¹²¹ This latter approach, which is in line with the vantage point of assessment of other elements of the plea, is also to be preferred for this requirement, as it judges the State by reference to what it could have known at the time and considering the projected effectiveness of the measures adopted, rather than what became known only ex post.

(iii) The problem of alternatives

The problem of alternatives involves a counterfactual: did the State have any alternative measures available to it? Investment tribunals have often been quick to accept that other alternatives are available and thus dismiss that the State’s challenged measure was the only way.¹²²

On paper, one can always find alternatives, especially with the benefit of hindsight, but this is not enough to dismiss the only way criterion.¹²³ Claimants submit expert reports, produced without the stress of an unfolding crisis and the pressure of governmental decision-making, and these inevitably identify other ways to tackle the crisis. But as Matthew Parish observed: a ‘general principle to the effect that any theoretical alternative means to the course pursued negates a defence of necessity, no matter how expensive, cannot be right. It would allow the defence to be defeated by wild, theoretical or whimsical suppositions.’¹²⁴

Where alternatives are available, they should defeat the ‘only way’ criterion only if they possess three features. First, an alternative measure must be lawful. Only the realistic availability of lawful alternatives disqualifies the State from invoking necessity. However, the Enron annulment committee suggested where other unlawful alternatives are available, to succeed in invoking necessity the State ought to take the least harmful alternative towards third-party rights.¹²⁵ While this may seem reasonable or even desirable,¹²⁶ the defence of necessity does not require it. The availability of alternative unlawful means does not negate the only way criterion. The defence permits causing a lesser harm, but it does not require causing the least harm.

A requirement to take least-harmful means could, potentially, be applied to the defence of necessity if it was considered that, in these situations, the reasons against the conduct in question are weighty and continue to operate in the background despite the presence of necessity. If, that is, all-things-considered the reasons against action have been overridden, but not canceled, by the situation of necessity. Where they are overridden, these reasons remain operative (in the background) and as such they continue to ‘exert their rational appeal’ (in John Gardner’s words). In these cases, the continued background operation of the reasons against the conduct leave a moral remainder, the implications of which would include an obligation to adopt the least-harmful means possible¹²⁷

Second, an alternative measure must be feasible in the sense that it is available to the relevant State at the time.¹²⁸ For example, States could build temporary hospitals to reduce pressure on existing healthcare structures during the SARS-CoV-2 pandemic. This could be a more costly and perhaps less convenient, yet lawful, way to protect lives and the healthcare system than imposing strict lockdowns and wide-ranging business closures. However, it may still be infeasible as the emergency may require an urgent response and there may be too few medical professionals available to staff these additional facilities. For a measure to be feasible, it must also be one that does not cause significant harm to the very interest that the State seeks to protect.¹²⁹ Since the purpose of the plea is to protect essential interests,¹³⁰ it would be incompatible with this aim if the essential interest sought to be protected was significantly harmed. Thus, for example, in the context of a public health emergency any lawful measure that involved risks to many more lives than the necessity measure cannot be a feasible alternative—precisely when the State’s measures are aimed at protecting those lives.

Third, an alternative measure, like the chosen measure, must be effective in the sense that it can safeguard the essential interest under threat. In respect of this last feature, Manton favours a threshold test: the measure either can, or it cannot, protect the relevant interest.¹³¹ If there is any other measure which can protect the interest (and is lawful and feasible), this suffices to deny the only way criterion. For the Enron Committee, instead, tribunals may need to engage in a comparative analysis of the effectiveness of the two (or more) measures: which one would have been more effective at protecting the interest?¹³² This requires assessing, at the time the State took the measures, the projected effectiveness of that measure in safeguarding the relevant interest. It follows from the very rationale of the plea, the protection of superior interests, that whichever measure was estimated, objectively, to offer the highest protection/ensure the least harm to the interest in question should be deemed the only way. That is, the identification of other lawful and feasible measures which would have been less effective in protecting the relevant interest than the measure in fact adopted does not exclude the latter as the ‘only way’. To be sure, comparisons between projected effectiveness of measures are likely to be complex, particularly under uncertainty. This complexity offers no good reason to reject comparisons altogether, since comparisons are consonant with the rationale of the plea that the interest under threat be safeguarded as much as possible.¹³³ An alternative measure which was lawful and feasible, but which contributed only minimally to the protection of the relevant interest could not displace the unlawful measure that was predicted to be more successful.

In sum, with respect to the only way criterion, tribunals ought to take the standpoint of the invoking State, by reference to evidence reasonably available to the State at the time it adopted the measures of necessity. Further, tribunals ought to dismiss other measures, even if lawful and feasible for the State, if they are less likely to protect, or less good at protecting, the interest in question, and ought to be conscious of hindsight bias.

D. Non-contribution

The plea of necessity is excluded if the State has contributed to the situation of necessity, namely to the risk of grave and imminent harm to its essential interest. The requirement incentivizes States to act prudently, and to avoid their putting their own interests excessively at risk. It prevents moral hazard.¹³⁴ The requirement of non-contribution, unlike the other requirements discussed above, lacks a clear basis in customary law: before its inclusion in the ARS, it had not featured in State practice concerning necessity.¹³⁵ Moreover, States did not unanimously support its inclusion in the ARS.¹³⁶ The non-contribution requirement is also vague and has given rise to significant interpretive difficulties in practice.

First, it is unclear what amounts to ‘substantial’ contribution: the ARS only state that it is more than ‘peripheral’ and tribunals have not addressed, developed or explained what ‘substantial’ means.¹³⁷ Investment tribunals have tended to list the State acts or omissions which contributed to the situation of necessity and then leap to the conclusion that the contribution was substantial (or not, as the case may be). For example, in CMS, the Tribunal stated that ‘government policies and their shortcomings significantly contributed to the crisis and the emergency and while exogenous factors did fuel additional difficulties they do not exempt the Respondent from its responsibility in the matter’.¹³⁸ Tribunals tend to reach their conclusion on this question, that is, from the minor premise only (the available facts). In at least one instance the award was annulled precisely because the tribunal had missed one step in the reasoning: the explanation relating to the major premise (the legal standard of ‘substantial contribution’) of the legal syllogism.¹³⁹

Second, it is not clear whether non-contribution is a purely causal requirement, in the sense that any contribution, if substantial, excludes the defence, or one requiring some degree of fault. The Commentary suggest a purely causal understanding,¹⁴⁰ and some investment awards have endorsed this approach, explicitly or implicitly.¹⁴¹ Other tribunals have, instead, adopted a fault-based conception of ‘substantial contribution’ and thus required that the contributing conduct be—at the very least—negligent.¹⁴² Several scholars support this latter understanding,¹⁴³ and the ILC preferred it on first reading.¹⁴⁴ Third, the temporal dimension of this requirement is uncertain: how far back in time can we go in search of contributing causes before these are judged too remote?¹⁴⁵

A very strict reading of this requirement—such as a purely causal reading—likely makes the plea unarguable. In most cases where a situation of necessity arises, there will be some conduct of the invoking State which may be said to have contributed to it—be it economic crises, wars or civil strife.¹⁴⁶ Robert Sloane, for example, has noted that in a ‘dynamic legal global order of increasingly interdependent states’ most situations of necessity will be caused by a combination of factors endogenous and exogenous to the invoking State.¹⁴⁷ The Unión Fenosa award echoed this view: ‘To an extent, a situation of necessity can always be traced back, as a matter of history, to political and economic mistakes made by a State years, if not decades, earlier.’¹⁴⁸ The Argentine financial crisis cases bear this out too: in almost all cases, the plea failed, among others, as tribunals found that Argentina’s policies had contributed to the crisis.¹⁴⁹ More troublingly, the requirement also leads to illogical or undesirable results.¹⁵⁰ If the point of the plea is to safeguard essential interests, then is it really relevant that the State contributed to the situation threatening those interests if the situation is one of real danger? During the ILC’s discussion on this condition, Roberto Ago highlighted that:

it was a moot point whether a State was in a ‘state of necessity’ if it found itself in a situation which was not really ‘caused by’ it but to which it had contributed, for instance by pursuing too lax a financial policy. If that situation was nevertheless fraught with extreme danger for it, was it fair not to allow it any excuse?¹⁵¹

In light of this, a fault-based standard would be preferable. This could be based on negligence, so that only negligent substantial contributions exclude the plea.¹⁵² At the same time, it must not be forgotten that in the past, States abused the defence of necessity and cognate concepts, like self-preservation.¹⁵³ A broader standard of non-contribution leaves some space for a defence of necessity despite risky behaviour by States. The prevention of moral hazard is an important goal, but not the only or the overriding goal, of necessity. Ago’s statement in the ILC echoes general criticisms of moral hazard, based among others on the moral responsibility to alleviate the suffering of others.¹⁵⁴ Moral hazard within the plea of necessity can also be prevented through different means: a duty of compensation for material loss caused by acts of necessity could, for example, accompany a more lenient non-contribution requirement.¹⁵⁵

E. The Temporal Dimension: The Duration of the Situation of Necessity

Article 27(a) indicates that the successful invocation of any defence is without prejudice to ‘compliance with the obligation in question, if and to the extent that the circumstance precluding wrongfulness no longer exists’. That is, States need to resume compliance with the relevant obligation as soon as the defence of necessity ceases to apply. This is a consequence of the generally accepted principle that necessity—like the other defences—does not ‘as such affect the underlying obligation’.¹⁵⁶ Where the obligation requires continuing performance, the state of necessity operates to set aside the obligation to comply while the situation of necessity lasts. However, once the necessity comes to an end, the State needs to resume compliance with the obligation. If the State fails to do so, it is pro tanto responsible for violations commencing after the end of the situation of necessity.¹⁵⁷

Determining the start and end dates of the situation of necessity can be a difficult and imprecise assessment, especially in situations of crises or emergencies. These do not need to overlap with the situation of necessity. This is a matter that requires objective determination, and one, therefore, that adjudicators need to make when deciding the plea.

As to the start of the situation of necessity, this involves determining the moment when a risk of grave and imminent harm to an essential interest arose. The defence is only triggered at the point (usually in a continuum) when an ongoing crisis begins to pose a risk of grave harm to the essential interest. With respect to the Argentine financial crisis, the LG&E Tribunal considered that the situation of necessity had commenced on 1 December 2001, the date when the government enacted the so-called Corralito.¹⁵⁸ The Impregilo Tribunal, by contrast, found that it commenced at the end of 2001—without pinpointing a specific date. By this time, according to the Tribunal, the Corralito, the ensuing demonstrations and deaths, the resignation of President de la Rúa, the succession of five presidents over the next 10 days, and the sovereign default on external debt showed that the situation had become critical.¹⁵⁹ In respect of the Arab Spring, the Tribunal in Unión Fenosa found that had there been a risk of grave and imminent harm to an essential interest, the situation of necessity would have started in ‘the spring of 2011 (when President Mubarak resigned from office)’.¹⁶⁰

The situation of necessity presumably ends when there is no longer a risk of imminent and grave harm to an essential interest. This can occur when the harm materializes and there is nothing left for the State to do other than deal with its consequences, or when successful State measures have partly or fully averted the materialization of the harm. The Commentary does not say this, but the endpoint of the situation of necessity is also for tribunals to determine objectively. This can be an imprecise assessment, and one about which tribunals may and do differ. The Unión Fenosa Tribunal considered that the situation of necessity would have ended ‘by 2015’ when ‘the Egyptian revolution had ceased’.¹⁶¹ On the Argentine crisis, the LG&E Tribunal determined that the state of necessity ended on 26 April 2003 with the election of President Kirchner,¹⁶² as at this point in time the ‘situation ha[d] been overcome, i.e. a certain degree of stability ha[d] been recovered’.¹⁶³ The Tribunal in Hochtief agreed that Argentina’s crisis had ended by mid-2003 and set the date for the end of the state of necessity on 1 May 2003, considering ‘the practicalities of financial accounting and calculation’.¹⁶⁴ For the SAUR Tribunal, the state of necessity came to an end in 2005, the date when Argentina was able to restructure its foreign debt and exit default.¹⁶⁵

It is not necessary that the situation return to its pre-crisis normality before the state of necessity ends. The Hochtief Tribunal was conscious of this and noted that despite having determined that the situation of necessity had ended by mid-2003, as economic growth had resumed by this point and in macroeconomic terms the worst of the crisis was over, ‘the social and other effects on Argentine society … persisted well after that time’.¹⁶⁶ In some instances, moreover, crises may affect the baseline of normality, such that a return to the pre-crisis normality may be impossible. The SARS-CoV-2 pandemic is one of the cases where a return to the pre-crisis normality (namely, where SARS-CoV-2 has been suppressed and no remaining human reservoirs remain in the world) is unlikely.¹⁶⁷ The new baseline is one where the virus becomes endemic, meaning that a certain amount of disease is a constant presence in a population in a particular geographical area.¹⁶⁸ Determining what the level of this new normal is, such that public health and related measures can be lifted, requires governments assessing what level of disease and mortality they, and their societies, are willing to accept.

In the absence of clear criteria to determine the duration of the situation of necessity, it is for adjudicators to stipulate when the state of necessity begins and when it ends. As the Tribunal in Hochtief explained this can be a ‘highly subjective exercise’, as it is one ‘overwhelmingly influenced by the precise factors or indicia upon which one focuses and the degree of change or stability that one regards as qualifying as a return to a normal, non-crisis situation’.¹⁶⁹ But it is a necessary exercise in order to ‘give effect to the tribunal’s reasoned judgment’.¹⁷⁰ All that can be expected of tribunals in these circumstances is, indeed, ‘a serious attempt by a tribunal to identify a suitable date on the basis of the evidence provided by the Parties’.¹⁷¹

Lastly, the end of crises and emergencies may be patchy and uneven.¹⁷² The Commentary to Article 27(a) envisages situations in which ‘the conditions preventing compliance gradually lessen and allow for partial performance of the obligation’.¹⁷³ In situations of necessity, this suggests that States are partially justified for failing to fully perform their obligation. This is not so, however. At a theoretical level, at least in international law, justifications (unlike excuses) cannot be partial: conduct is either justified, in which case it is lawful, or it is not, in which case it is unlawful.¹⁷⁴ In practice, moreover, the Commentary to Article 12 explicitly excludes this possibility. It states that a breach of international law ‘may exist even if the act of the State is only partly contrary to an international obligation incumbent upon it’.¹⁷⁵ Partial performance amounts to a breach of the State’s international obligations. The extent of the breach, however, affects reparation due and the assessment of damages for purposes of compensation.

F. Duty of Compensation for Material Loss

Article 27(b) also contains a without prejudice clause in respect of compensation for material loss caused by acts covered by defences.¹⁷⁶ The duty of compensation envisaged in Article 27 is beset by two main ambiguities. First, its legal basis is unclear. The Commentary clarifies that this basis differs from the one for compensation as a form of reparation,¹⁷⁷ perhaps suggesting that compensation is not due to the wrongful character of the act. At the same time, the drafting of this provision shows that the Commission dismissed this duty as a form of liability arising from lawful conduct.¹⁷⁸ Second, it is unclear in respect of which defences it applies and whether it applies to every instance of the successful invocation of certain defences. The Commentary nevertheless suggests a moral hazard rationale for this duty: ‘Without the possibility of such recourse, the State whose conduct would otherwise be unlawful might seek to shift the burden of the defence of its own interests or concerns onto an innocent third State.’¹⁷⁹

Many investment awards discuss the duty of compensation—unsurprisingly, as investors seek first and foremost to recover their losses. For the most part, tribunals’ observations on this duty have been in obiter: no public award has, to date, awarded compensation for conduct justified by necessity. Their engagement with this duty has, nevertheless, been far from satisfactory. Tribunals in CMS, BG and EDF endorsed the existence of a duty of compensation in cases of necessity. The CMS Tribunal relied on past decided cases and general principles,¹⁸⁰ the BG Tribunal grounded the duty on the language of the relevant BIT,¹⁸¹ and the EDF Tribunal and Committee inferred this duty from the concept of necessity.¹⁸² The SAS v Bolivia Tribunal simply asserted the existence of this duty.¹⁸³ Other tribunals have been more sceptical—noting that Article 27(b) is a without prejudice clause and not a stipulation.¹⁸⁴ Others still, like Enron and Sempra, noted that Article 27(b) was vague. They held in line with the Commentary that the parties should decide whether compensation is due, and added that absent their agreement ‘this determination is to be made by the Tribunal to which the dispute has been submitted’.¹⁸⁵

The case law has not moved the debate any further. No tribunal has seriously engaged with the positive law status of this duty, in respect of which scholarly opinions differ.¹⁸⁶ Many scholars agree that a duty of compensation in these circumstances is desirable,¹⁸⁷ but so far States have rarely expressed their views on this duty and their views are mixed.¹⁸⁸ Insofar as the duty to compensate is a mechanism for the reduction of moral hazard, it could be used as a trade-off for a more relaxed reading of certain other requirements of the plea sharing the same rationale, such as the requirement of non-contribution as suggested earlier. As a solution of the question whether compensation should accompany a successful invocation of the plea of necessity would require a modification of Article 25, this is for States to decide—just like their decision to include internal exceptions in response to necessity’s shortcomings, to which we now turn.

III. LESSONS FROM THE INVESTMENT CASE LAW ON NECESSITY

The term ‘necessity’ has many meanings.¹⁸⁹ In ordinary language, the meaning of necessity refers, among others to situations of want or deprivation, and situations of hardship or difficulty.¹⁹⁰ Sarah Heathcote refers to these senses of necessity as ‘the necessity of predicament’.¹⁹¹ In this meaning, necessity plays different functions in the law.¹⁹² These functions are often expressed with the medieval canon law maxims necessitas non habet legem (necessity has no law) and necessitas facit legem (necessity makes law).¹⁹³ Neither of these maxims expresses legal rules: the existence of a situation of necessity does not generate rights for individuals, nor does it open the door to non-legality.¹⁹⁴ As the Umpire in the Faber case noted: ‘[t]he rights of an individual are not created or determined by his wants or even his necessities’.¹⁹⁵ The same is true of States. These maxims refer to the ordinary concept of necessity and how it interacts with legal rules: by highlighting areas of absent or insufficient regulation (it has no law), and by inspiring the creation or modification of rules of law (it makes the law).¹⁹⁶ In other words, necessity is not a formal source of law, but it can be a material source.¹⁹⁷ A clear example of this function of necessity is the Convention Relating to Intervention on the High Seas in Cases of Oil Pollution Casualties,¹⁹⁸ adopted in 1969 in response to the Torrey Canyon incident of 1967.¹⁹⁹ Other rules inspired by necessity include the right of self-defence, the right of angary and general defences of necessity, like Article 25 ARS.

There is broad agreement in the literature that the main lesson that can be drawn from the investment arbitration experience with the necessity defence is that this defence did not work for the types of crises addressed by investment tribunals, in particular Argentina’s financial crisis. Nevertheless, the predicament that Argentina found itself in during its financial crisis, and its inability to rely on the defence of necessity, highlighted how limited was the regulatory space that investment treaties, absent (significant) internal exceptions, safeguarded for States. But the impact of necessity may be felt elsewhere in investment law beyond adjudication: unable to work as a legal rule, necessity in its ordinary meaning highlighted a gap in the existing regulation and led lawmakers (States) to modify those, or create new, rules. Indeed, this recognition may partly explain the move by States towards inclusion of (internal) exceptions in their investment treaties (as well as reform of investment protection standards). Until a decade ago, fewer than 10 per cent of investment treaties contained some form of internal exceptions.²⁰⁰ Albeit from a low base, States have increasingly included internal exceptions in their investment treaties in the decade 2010–20.²⁰¹

These internal exceptions in investment treaties have taken various forms by reference to the interests they seek to safeguard. Also, by reference to those interests, States have adopted different standards (or ‘nexus requirements’) between the permitted goal (safeguard a certain interest) and the permitted measures.²⁰² Among others, these measures have included: general exceptions modelled on GATT Article XX; non-precluded measures provisions that were a prominent feature of the arbitrations arising out of Argentina’s financial crisis based on Article XI of the Argentina–US BIT; essential security clauses inspired by GATT Article XXI; and as regards public health, general public health exceptions as well as specific disease exceptions. Beyond the Argentine financial crisis cases, practice on internal exceptions under investment treaties is so far limited.²⁰³

As an example of this shift in investment treaty-making practices, the increase in public health exceptions in investment treaties since 2016 has been around 30 percentage points (excluding investment treaties that refer to public health only in the preamble).²⁰⁴ Of 121 investment treaties concluded since 2016, circa 42 per cent include some form of public health exception, compared to around 10 per cent (250 treaties) out of around 2600 treaties concluded over 1959–2015.²⁰⁵ This is, of course, a gradual trend because renegotiating investment treaties and revising model investment treaties tends to take years.²⁰⁶ Nevertheless, the increase in internal exceptions is substantial. Table 1 contrasts the nexus requirements of necessity with four types of internal exceptions increasingly found in investment treaties.

Table 1.

Comparison of necessity and internal exceptions in investment treaties

	Necessity	General exceptions	Essential security	Public health exception	Disease exception1
Nexus requirement	Very strict	Strict	Very lenient	Lenient	Lenient
Nexus	Only way criterion	‘necessary to’	‘relate to’	‘directed to’	‘directed to’
Per cent of IIAs (total)	Custom	32	16	10	1.6
Per cent of IIAs (since 2016)	Custom	85	27	42	None (last BIT 2014)

Source: Authors’ calculations, drawing on investment treaties available in English in UNCTAD’s IIA database.

Notes  ¹Investment treaties that include an exception concerning ‘pest’, ‘pests’ or ‘diseases’. Three countries are parties to 44 of the 46 investment treaties concerned (India, Mauritius and Singapore).

2

Wolfgang Alschner and Kun Hui, ‘Missing in Action: General Public Policy Exceptions in Investment Treaties’ in Lisa Sachs and others (eds), 2018 Yearbook on International Investment Law and Policy (OUP 2019) (based on the Electronic Database of Investment Treaties).

The requirements for all four types of internal exceptions are less stringent, to varying degrees, than those that apply to necessity. The first alternative to necessity are general exceptions modelled on GATT Article XX, in particular measures ‘necessary to protect human, animal or plant life or health’. These exceptions have a broad scope, but like necessity are subject to a strict nexus requirement. Depending on how investment tribunals interpret these exceptions, host states may not preserve more regulatory space by including general exceptions in the investment treaty compared to customary international law.²⁰⁷

Second, essential security exceptions, with their typically self-judging language, enable States to invoke them in emergencies subject to few, if any, limitations. Even if international adjudicators can review the invocation of essential security exceptions, as Ben Heath explains, ‘the discretion afforded by the phrase “[the State] considers”, combined with the indeterminacy of other undefined terms, suggests that States retain wide discretion in security matters’.²⁰⁸ Essential security exceptions are subject only to a very lenient nexus requirement. Investment tribunals have analogized imminent threat to a State’s economy as an essential security.²⁰⁹ A central factor whether a crisis qualifies as an essential security interest is the intensity of the threat,²¹⁰ which is a matter for the State to determine.

Third, public health exceptions (which partly overlap with general exceptions) are exceptions with a lenient nexus requirement (a rational relationship with public health). Measures that are rationally related to public health, based on scientific evidence, ‘shift[] the risk of state action to investors in the case of exceptional actions taken to protect public health’.²¹¹ Because of the exceptions’ lenient nexus requirement and because States can rely on scientific evidence, States may find it easier to invoke public health exceptions than to invoke necessity. That said, public health exceptions have not yet played a significant role as a defence in investor-State arbitration,²¹² and at present, much remains unknown about their operation.

Fourth, specific disease exceptions are narrow exceptions which vary in language: some specific disease exceptions refer to the ‘prevention of diseases and pests’²¹³ (and not public health more generally); others refer to ‘the protection of public health or the prevention of diseases or pests in animals or plants’,²¹⁴ and thus combine the third variant with a specific variant focused on diseases. Still others only refer to ‘the prevention of diseases and pests in animals and plants’, but not to public health, and conceivably exclude human health from the ambit of the exception.²¹⁵ the SARS-CoV-2 pandemic might affect exceptions in investment treaties including public health exceptions going forward,²¹⁶ particularly if the customary defence of necessity turns out not to cover the measures adopted in response to the pandemic.²¹⁷

Essential security exceptions, public health exceptions and specific disease exceptions respond to the legitimacy problem of the defence of necessity. States as lawmakers weight the relevant interests ex ante, rather than adjudicators having to weigh these interests ex post. Moreover, from the perspective of regulatory space, the value added of internal exceptions is the relaxed nexus requirement (‘related to’ or ‘directed to’)—more lenient than general exceptions, and much more lenient than the only way criterion for the customary defence of necessity. Measures could pass muster if they further the permissible objective of protecting public health.²¹⁸ In Devas v India, the Tribunal considered that to the extent that the reservation of spectrum was directed towards India’s military needs, it fell within India’s ‘essential security interests’.²¹⁹ By contrast, spectrum reserved for non-military uses fell within the ambit of the investment treaty.

That said, the shift to internal exceptions could also have the unintended effect of making it more difficult for States to rely on the regulatory purpose of measures when defending them. Future investment tribunals could be less inclined to take the regulatory aim of State measures into account at the level of the treatment standard, particularly regarding fair and equitable treatment, indirect expropriation and national treatment,²²⁰ unless the investment treaty expressly directed investment tribunals to not regard non-discriminatory regulatory measures as an indirect expropriation and so on.²²¹ Tribunals might interpret treaties not containing exceptions as not allowing any, as opposed to other treaties. Exhaustively drafted exceptions might undermine States’ regulatory priorities.²²² Tribunals must therefore guard against such interpretations.

IV. CONCLUSION

Article 25 is the first time that the defence of necessity in international law was fully written down in an instrument endorsed by States. The case law and practice on which the ILC drew for Article 25 was supportive of the principle expressed in this rule, but it had not specified precise conditions for necessity’s application.²²³  Gabčíkovo–Nagymaros gave a teaser of the difficulties ahead in the formulation and implementation of this rule. Indeed, in reaction to this judgment, the ILC adapted some elements of this provision. But the new formulation of the rule in Article 25 was yet to be tested at the time of ARSIWA’s conclusion in 2001. It was not until the investment case law of the early twenty-first century, dealing primarily with the Argentine financial crisis, that the difficulties in the drafting of Article 25 and, indeed, in the concept of necessity itself began to reveal themselves.

A first difficulty relates to the vantage point of assessment of claims of necessity, a matter which the ILC did not clearly address in the Commentary. To be sure, invocations of necessity must be assessed objectively, but this says nothing as to the vantage point: objectively from the point of view of the tribunal, with its hindsight knowledge, or objectively from the vantage point of the State at the time of acting? Allowing tribunals to judge these invocations with the benefit of hindsight can lead to illogical or, at the very least, unfair conclusions. A tribunal assessing the plea from this standpoint does so with a ‘God’s eye perspective’: only God knows how things turn out and, from this vantage point, the ex ante perspective (that of the State at the time of acting in necessity) and the ex post perspective (that of the adjudicators) are identical. Yet when States act in necessity they do not know (they cannot know) how things will turn out. The threshold for the defence ranges from high, to impossible to meet at worst, or a matter of luck at best. Investment tribunals did not always show an appreciation of this problem. But even if they had, there are inherent difficulties in assessing the claim from the vantage point of the State: the tribunal knows how things turned out, and it cannot unknow this. This knowledge can, and indeed did, lead to hindsight bias. When assessing necessity pleas, adjudicators need to be especially careful to mitigate hindsight bias.

The second difficulty is associated with the standard of proof required to demonstrate the existence of a grave and imminent peril (both the triggering event and the expected harm). The harm to the State’s essential interest is in the future, and States can never prove the occurrence with certainty. The ILC allowed some space for the invocation of necessity in situations of risk, accepting that some risk does not preclude establishing the peril. However, it still failed to provide a clear standard of proof (balance of probabilities, lower or higher?) and the investment case law has not engaged with this question in any detail.

Third, the only way criterion is not only unduly strict but also wholly unrealistic for macro crises, as demonstrated by the Argentine cases in respect of financial crises. A State usually has multiple options in addressing a crisis. Moreover, when addressing such macro crises, States may need to adopt packages of measures, and the effectiveness of these measures is rarely known ex ante. States may estimate the effectiveness of individual measures and packages of measures, but this is a prediction: some measures work, others work partially and still others fail. The vantage point of assessment becomes critical again in this respect: we can only really know if a measure was the only way to deal with a crisis once we know if it has worked. But the State cannot know this in advance. The requirement is not only difficult but impossible to meet.

The ILC had good reasons to impose a high threshold for the invocation of this (and indeed, any) defence in international law: after all, these are invoked when there are allegations of breach of international law and the affected State usually has no binding mechanisms to challenge these invocations. Nevertheless, the investment law experience with this defence has proven right those ILC members who warned that such a tightly drafted defence would ‘deprive it of its usefulness altogether’.²²⁴ It may be that the ILC intended to exclude reliance on the plea in situations involving macro-crises, where many interests are in conflict. Perhaps the ILC desired that these crises be resolved through cooperation, but the feasibility of cooperative solutions depends on several contingent factors, including time and States’ relative economic and political weight. Moreover, as formulated, the plea of necessity is a blunt instrument: it is an all-or-nothing rule, the conduct is either permitted or it is prohibited, and it does not allow for adjustment.

The defence of necessity has not provided the promised ‘safety valve’²²⁵ to States like Argentina when facing serious upheaval and instability. The clinical narrative of events in awards, in academic articles, and indeed in the courtroom cannot do justice to the realities on the ground such as demonstrations, deaths and the succession of five presidents over the next 10 days.²²⁵ The feverish atmosphere and volatility of these situations, to which those who have lived through them and experienced them first-hand can attest, is difficult to convey with words. Argentina needed a safety valve, whether necessity or some other rule. Even though the detached perspective of adjudicators can be valuable, it can give rise to hindsight bias (which may be impossible to completely overcome) and provide a justification for adjudicators to defer to the State invoking necessity in certain contexts.

If necessity has not worked as a legal rule, the concept of necessity in its ordinary meaning, one denoting situations of difficulty or predicament, has instead performed one of its functions with respect to the international legal order: financial crises and pandemics are situations of urgent need, in the ordinary sense of the word. The Argentine crisis highlighted just how little regulatory space had been left to States by investment treaties (particularly their treatment standards) and has nudged States towards reconsidering the substantive obligations in their investment treaties and to include internal exceptions. These internal exceptions cater to situations of pressing need, of danger to essential interests, but they differ in that they specify the interests for which they are available in advance and provide for more regulatory space for States than necessity. The defence of necessity has had little purchase in investor-State arbitrations, but the real necessity performed its function: it has contributed to ‘making’ international investment law.

Footnotes

1

Federica Paddeu, Derek Bowett Fellow, Queens’ College, Cambridge, United Kingdom. Email: [email protected]. Michael Waibel, Professor of International Law, University of Vienna, Austria. Email: [email protected]. We thank Damien Charlotin, Laura Diaz Anadon, Caroline Foster, J Benton Heath, Caroline Henckels, Ko-Yung Tung and the peer reviewers for comments, and Isabella Brunner and Peter Behyl for research and editorial assistance.

2

As regards investment, it would be far too demanding to ask the doctrine of necessity to do most or all of the work in safeguarding regulatory space in economic or public health crises. This task primarily falls to the treatment standards such as fair and equitable treatment and expropriation, including deference to host States under these standards.

3

James Crawford, ‘The International Court of Justice and the Law of State Responsibility’ in Christian J Tams and James Sloane (eds), The Development of International Law by the International Court of Justice (OUP 2013) 74. Codification involves a degree of progressive development: see Mark E Villiger, Customary International Law and Treaties: A Study of Their Interactions and Interrelations, with Special Consideration of the 1969 Vienna Convention on the Law of Treaties (Brill 1985) 126.

4

This definition closely tracks what Sarah Heathcote called the ‘received’ definition of necessity: Heathcote, ‘State of Necessity and International Law’ (PhD thesis, Graduate Institute of International and Development Studies, 2005, on file with the authors) ch 1. Nevertheless, the specific formulation of the various elements (a situation of danger, conflicting interests, necessity of the act and non-contribution) changes: general rules of necessity contain these elements, but their specification differs.

5

Beyond investment arbitration, the only court or tribunal to have considered necessity is the International Court of Justice in Palestinian Wall: Legal Consequences of the Construction of a Wall in the Occupied Palestinian Territory [2004] 43 ILM 1009 (2004) Advisory Opinion 1009. Many States commented on the plea in their submissions to the Court, but the Court addressed it in a short paragraph: para 140.

6

Guris Construction and Engineering Inc v Syrian Arab Republic, ICC Case No 21845/ZF/AYZ, Award (31 August 2020) (necessity invoked in relation to the armed conflict in Syria).

7

Unión Fenosa Gas SA v Arab Republic of Egypt, ICSID Case No ARB/14/4, Award (31 August 2018) (the Tribunal considered that the Respondent’s measures and the gas shortage were causally unrelated to the revolution).

8

JKX Oil & Gas plc, Poltava Gas BV and Poltava Petroleum Company JV v Ukraine, UNCITRAL, Award (6 February 2017) (unpublished). Ukraine invoked necessity in relation to the crisis in Crimea. The Tribunal denied the defence on the grounds that the Respondent’s measures were not the ‘only way’: Damien Charlotin, ‘Revealed: Tribunal in JKX v Ukraine Awarded Nearly 12 Million USD for Arbitrary Measures and Breach of Free Transfer Clause; Ukraine’s Necessity Defence Was Rejected’, IAReporter, (29 June 2020) <www.iareporter.com/articles/on-the-wires-philippines-liable-for-bit-breach-investor-that-won-emergency-orders-vs-ukraine-does-not-fare-as-well-in-final-result-iranians-pursue-bahrain-over-bank-closure/> accessed 12 march 2021.

9

Bernhard von Pezold and Others v Republic of Zimbabwe, ICSID Case No ARB/10/15, Award (28 July 2015) (necessity invoked in respect of land occupations and ensuing public protests; the Tribunal dismissed the plea on various grounds); South American Silver v Bolivia, PCA Case No 2013–15, Award (22 November 2018) (necessity plea invoked in relation to indigenous peoples’ protest; the Tribunal found that necessity did not apply to expropriation claims).

10

Larry Alexander, ‘Lesser Evils: A Closer Look at the Paradigmatic Justification’ (2005) 24 Law & Philosophy 611.

11

eg CMS Gas Transmission Company v Argentine Republic, ICSID Case No ARB/01/8, Annulment (25 September 2007) paras 129–34; Continental Casualty Company v Argentine Republic, ICSID Case No ARB/03/9, Award (5 September 2008) fn 236.

12

Many scholars have argued in favour of the recognition of excuses and, in particular, the characterization of the defence of necessity as an excuse: Rosalyn Higgins, Problems and Process: International Law and How We Use It (Clarendon Press 1994) 40; Ian Johnstone, ‘The Plea of “Necessity” in International Legal Discourse: Humanitarian Intervention and Counter-terrorism’ (2005) 43 Columbia J Transl L 337, 352–56; Theodore Christakis, “Nécessité n’a pas de Loi”? Rapport général sur la nécessité en droit international’ in Theodore Christakis (ed), La nécessité en droit international (Pedone 2007) 47ff; Francisco Orrego-Vicuña, ‘Softening Necessity’ in Mahnoush H Arsanjani (ed), Looking to the Future: Essays on International Law in Honour of W Michael Reisman (Brill 2010) 742; Cedric Ryngaert, ‘State Responsibility, Necessity and Human Rights’ (2010) 41 NYIL 79, 96–97; Federica Paddeu, Justification and Excuse in International Law: Concept and Theory of General Defences (CUP 2018) 426–29.

13

Sempra Energy International v Argentine Republic, ICSID Case No ARB/02/16, Award (29 September 2007) para 333 (a ‘concept precluding wrongfulness’); Enron Corporation and Ponderosa Assets LP v Argentine Republic, ICSID Case No ARB/01/3, Award (22 May 2007) para 294 (a ‘concept precluding wrongfulness’); Continental Casualty (n 11) para 160 (actions ‘entirely lawful’); Metalpar SA and Buen Aire SA v Argentine Republic, ICSID Case No ARB/03/5, Award (6 June 2008) para 137 (‘justification … under international law’); National Grid plc v The Argentine Republic, UNCITRAL, Award (3 November 2008) para 205 (conduct is ‘licit’); EDF International SA, SAUR International SA and Leon Participaciones Argentinas SA v Argentine Republic, ICSID Case No ARB/03/23, Award (11 June 2012) para 1163 (conduct is ‘justified’); EDF International SA, SAUR International SA and Leon Participaciones Argentinas SA v Argentine Republic, ICSID Case No ARB/03/23, (5 February 2016) para 317 (plea invoked in case of finding of a ‘prima facie breach of the BIT’); Total SA v Argentine Republic, ICSID Case No ARB/04/1, Annulment (1 February 2016) para 229 (plea invoked as a ‘circumstance that precludes the wrongfulness of an act contrary to the obligation allegedly violated’). The pleadings of Argentina are not public, so the following list of examples relies on the Tribunal’s own summary of the Argentine argument. Even though Argentina’s position on the plea of necessity under customary law may have been consistent throughout the various disputes brought against it, Argentina drew a distinction between necessity and the non-precluded measures provisions in Article XI of the Argentina–US BIT only from Sempra onwards. Nevertheless, tribunals have used inconsistent and equivocal language in their description of Argentina’s argument, sometimes using the language of justification and other times the language of excuse. Troublingly, they have often done so within the same award.

14

South American Silver v Bolivia (n 9) para 534.

15

Unión Fenosa Gas SA v Arab Republic of Egypt (n 7) para 8.21: ‘[T]he state of necessity caused by the revolution that erupted in Egypt and whose after-effects persist to this day precludes any wrongfulness of the conduct alleged to constitute a violation of the Spain–Egypt BIT…’.

16

Von Pezold (n 9) para 616.

17

The position of Syria in Guris (n 6) is unclear: the award contains a few short paragraphs on state of necessity and force majeure, and uses the expressions ‘preclusion of wrongfulness’ and ‘exoneration of responsibility’ in respect of both (see paras 315–22).

18

See Paddeu (n 12) ch 8 for an extensive analysis of the practice supporting the lesser-evil rationale of the defence.

19

Economic models often assume full information (the absence of information asymmetries). Asymmetric information models relax this assumption. See eg Joseph Stiglitz, ‘The Contributions of the Economics of Information to Twentieth Century Economics’ (2000) 115 Quarterly Journal of Economics; Kenneth J Arrow, The Economics of Information (Harvard UP 1984); David De Meza and David C Webb, ‘Too Much Investment: A Problem of Asymmetric Information’ (1987) 102 Quarterly J Econ.

20

Jacqueline Peel, Science and Risk Regulation in International Law (CUP 2010) 85.

21

Expected harm refers to ‘the average harm that would occur were the act repeated many times’. See eg Steven Shavell, ‘The Optimal Structure of Law Enforcement’ (1993) 36 Journal of Law and Economics 256; Oren Bar-Gill and Gabriella Blum, ‘Defenses’ (2019) 97 Texas L Rev 881 (expected harm in the law of armed conflict).

22

See eg Michael Power, ‘The Risk Management of Everything’ (2004) Journal of Risk Finance; David A Moss, When All Else Fails: Government as the Ultimate Risk Manager (Harvard UP 2002); Robert J Shiller, The New Financial Order: Risk in the 21st Century (Princeton UP 2003).

23

Daniel Kahneman and others, Judgment under Uncertainty: Heuristics and Biases (CUP 1982) 3–28.

24

Peel (n 20) 86; Daniel A Färber, ‘Uncertainty’ (2011) 99 Geo LJ 901.

25

Frank H Knight, Risk, Uncertainty and Profit (Houghton Mifflin 1921); John Maynard Keynes, Treatise on Probability (Macmillan 1921). For a more recent discussion of risk and uncertainty, and tools for managing them in the context of climate change, see Howard Kunreuther and others, ‘Integrated Risk and Uncertainty Assessment of Climate Change Response Policies’ in Ottmar Edenhofer and others (eds), Climate Change 2014: Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (CUP 2014).

26

Usage of the term ‘uncertainty’ is not uniform, sometimes blurring the distinction. See eg David Spiegelhalter, ‘Risk and Uncertainty Communication’ (2017) 4 Annual Review of Statistics 31, 33 (dividing ‘true’ uncertainty into aleatory, epistemic and ontological kinds); Laura Díaz Anadón and others, ‘Integrating Uncertainty Into Public Energy Research and Development Decisions’ (2017) 2 Nature Energy (managing uncertainty on technological change).

27

Knight (n 25) 19.

28

Stephen C Nelson and Peter J Katzenstein, ‘Uncertainty, Risk, and the Financial Crisis of 2008’ (2014) 68 Intl Org 362.

29

John Kay and Mervyn King, ‘The Radical Uncertainties of Coronavirus’ (2020) Prospect Magazine <www.prospectmagazine.co.uk/magazine/coronavirus-model-uncertainty-kay-king>.

30

Kay and King (n 27).

31

On potential investment treaty claims in respect of States’ pandemic response, see Lucas Bento and Jingtian Chen, ‘Investment Treaty Claims in Pandemic Times: Potential Claims and Defense’ (Kluwer Arbitration Blog, 8 April 2020) <http://arbitrationblog.kluwerarbitration.com/2020/04/08/investment-treaty-claims-in-pandemic-times-potential-claims-and-defenses/>.

32

‘The Committee also acknowledged that there are still many unknowns’ <https://www.who.int/news/item/30-01-2020-statement-on-the-second-meeting-of-the-international-health-regulations-(2005)-emergency-committee-regarding-the-outbreak-of-novel-coronavirus-(2019-ncov)>; cf ‘Sir David Spiegelhalter: Risk Is a Very Loaded Term’ Financial Times (16 April 2021).

33

See eg D Sridhar and MS Majumder, ‘Modelling the Pandemic’ (2020) 369 BMJ (21 April 2020).

34

David Spiegelhalter and Anthony Masters, ‘How Good Are We at Predicting the Pandemic?’ The Guardian (9 May 20201) <www.theguardian.com/theobserver/commentisfree/2021/may/09/how-good-are-we-at-predicting-pandemic?>.

35

See eg, ARSIWA commentary art 25, para 15.

36

Jeffrey J Rachlinski, ‘A Positive Psychological Theory of Judging In Hindsight’ (1998) 65 U Chi L Rev.

37

Baruch Fischhoff, ‘Hindsight ≠ Foresight: The Effect of Outcome Knowledge on Judgment Under Certainty’ (1975) 1 J Experimental Psychology: Human Perception and Performance 288; Baruch Fischhoff, ‘For Those Condemned to Study Past: Heuristics and Biases in Hindsight’ in Kahneman and others (n 23); Hal R Arkes and others, ‘Eliminating the Hindsight Bias’ (1988) 73 Journal of Applied Psychology; Christine Jolls and others, ‘A Behavioral Approach to Law and Economics’ (1998) 50 Stan L Rev; Doron Teichman, ‘The Hindsight Bias and the Law in Hindsight’ in Eyal Zamir and Doron Teichman (eds), The Oxford Handbook of Behavioral Economics and the Law (OUP 2014).

38

Erin M Harley, ‘Hindsight Bias in Legal Decision Making’ (2007) 25 Social Cognition; Sergio Puig, ‘Debiasing International Economic Law’ (2019) 30 EJIL 1339, 1350 (‘practically irreversible’, also noting that unclear criteria may exacerbate the bias); Christine Jolls and Cass R Sunstein, ‘Debiasing through Law’ (2006) 35 JLS, table A1 (offering strategies for debiasing in general, but also noting the difficulties of correcting for hindsight bias specifically); Neal J Roese and Kathleen D Vohs, ‘Hindsight Bias’ (2012) 7 Perspectives on Psychological Science; Laura G Pedraza-Fariña and Ryan Whalen, ‘A Network Theory of Patentability’ (2020) 87 Uni Chi L Rev 71.

39

Oxford English Dictionary, ‘peril’, meaning 1(a).

40

Similarly, Sarah Cassella, La nécessité en droit international (Martinus Nijhoff 2011) 159.

41

Gabčíkovo–Nagymaros Project (Hungary/Slovakia) ICJ Reports 1997, 7, para 54: ‘The word “peril” certainly evokes the idea of “risk”; that is what distinguishes “peril” from material damage.’

42

Even though there is some work on causation in state responsibility in general, there is no literature on causation in the circumstances precluding wrongfulness, including necessity. See eg David Miklós Pusztai, ‘Causation in the Law of State Responsibility’ (PhD thesis, University of Cambridge, 2017, on file with the authors).

43

For an exploration, see Cassella (n 40) 159–64.

44

See eg, French Company of Venezuelan Railroads (1904) 10 RIAA 285.

45

In this sense, ‘imminent’ in the law of necessity is different from the same term when used in the law on the use of force: an ‘imminent’ armed attack is one that is about to happen. See Christian Henderson, The Use of Force and International Law (CUP 2018) 277ff. The standard of imminence goes back to the correspondence in the Caroline incident between the United Kingdom and the United States: see letter of Daniel Webster, US Secretary of State, to Lord Ashburton, special envoy of Great Britain to the United States, of 27 July 1842, in William R Manning (ed), Diplomatic Correspondence of the United States–Canadian Relations, vol 3 (Carnegie Endowment 1943) 145.

46

See ARSIWA commentary art 25, para 16; Gabčíkovo–Nagymaros (n 41) para 54. In the literature, see also James Crawford, State Responsibility: The General Part (CUP 2013) 310.

47

Where harm has begun to materialize, the State may still need to rely on forecasts of the possible extent and gravity of said harm.

48

Some theorists refer to this as God’s eye perspective: God knows how things turn out, and from God’s vantage point the ex ante and ex post perspectives are identical. See Mitchell N Berman, ‘Lesser Evils and Justification: A Less Close Look’ (2005) 24 L Phil 681, 703. Note, however, that uncertainty may also persist at the time of adjudication.

49

This is a well-known epistemic problem of consequentialism, and at the heart of the division between actual and expected consequentialisms. For an overview of the issue, see Walter Sinnott-Armstrong, ‘Consequentialism’ in EN Zalta (ed), Stanford Encyclopedia of Philosophy (2012) s 4.

50

Caroline E Foster, Science and Precautionary Principle in International Courts and Tribunal : Expert Evidence, Burden of Proof and Finality (CUP 2011) 18 (emphasis added).

51

The ILC discussed whether to include the precautionary principle into the concept of necessity, but chose to keep the two distinct: the plea of necessity is not intended to apply in situations of uncertainty. See: James Crawford, State Responsibility: The General Part (CUP 2013) 310–1.

52

ibid 6 (‘potentially decisive facts about future events clearly cannot be obtained at the time of adjudication’, emphasis added). On the limitations of the precautionary principle, see Eyal Benvenisti, Sharing Transboundary Resources: International Law and Optimal Resource Use (CUP 2002) 110 (‘The precautionary principle underscores the important role of risk management … But this principle is, at best, irrelevant in terms of determining just how, precisely, risks should be managed. It fails to define the level of uncertainty that triggers the principle into operation; who bears the onus of proving such uncertainty; and which principles should inform decision-makers in their cost/benefit analyses’); Daniel Bodansky, ‘Scientific Uncertainty and the Precautionary Principle’ (1991) 33 Environment: Science and Policy for Sustainable Development.

53

Foster (n 50) 18.

54

ARSIWA commentary art 25, para 15.

55

ibid.

56

Foster argues that the factual evidence was uncertain and should have led the Court to accept Hungary’s plea (Caroline E Foster, ‘Necessity and Precaution in International Law: Responding to Oblique Forms of Urgency’ (2008) 23 New Zealand Universities L Rev 265.

57

Gabčíkovo–Nagymaros (n 41) paras 55–56.

58

Paola Anna Pillitu, Lo stato di necessità nel diritto internazionale (Università di Perugia 1981) 302; Cassella (n 40) 169.

59

Note, though, that the UK did not invoke necessity in relation to this measure: Comments by Governments on all the Draft Articles, UN Doc A/CN.4/488 and Add.1–3, ILC YB 1998, vol II(1), 134–35.

60

See, in particular, in respect of the Nagymaros portion of the project: Gabčíkovo–Nagymaros (n 41) para 55. The plea failed because Hungary could not prove with the required degree of certainty that harm to the environment would occur, see Gabčíkovo–Nagymaros (n 41) paras 42ff.

61

See Subsection (ii) for more details on, and criticism of, this standard of proof.

62

See eg Metalpar (n 13) para 208; Impregilo Spa v Argentine Republic, ICSID Case No ARB/07/17, Award (21 June 2011) paras 347–50.

63

LG&E Energy Corp, LG&E Capital Corp, LG&E International Inc v Argentine Republic, ICSID Case No ARB/02/1, Decision on Liability (3 October 2006) paras 231–37.

64

CMS Gas Transmission Company v Argentine Republic, ICSID Case No ARB/01/8, Award (12 May 2005) paras 322, 354. Similarly, the tribunal in Sempra (n 13) required that the situation of necessity had already compromised the existence of the State (at para 348) or, as it stated in the context of necessity as a domestic law plea, the ‘constitutional order was not on the verge of collapse’ (at para 322).

65

The ICJ regularly holds that the evidence adduced by a party is ‘insufficient’ but refrains from specifying what the required standard of proof is. See Markus Benzing, ‘Evidentiary Issues’ in Andreas Zimmermann and others (eds), The Statute of the International Court of Justice: A Commentary (3rd edn, OUP 2019) para 108 (citing Corfu Channel (United Kingdom of Great Britain and Northern Ireland v Albania) (Judgment) [1949] ICJ Rep (1949) 4, 16; Military and Paramilitary Activities in and against Nicaragua (Nicaragua v United States of America) (Judgment) [1986], ICJ Rep, 14, 37, para 54, pp 62–63, para 110, and pp 85–86, para 159; Pulp Mills on the River Uruguay (Argentina v Uruguay) (Judgment) [2010] ICJ Rep 14, 97–98, para 254; Certain Activities Carried Out by Nicaragua in the Border Area (Costa Rica v Nicaragua) (Judgment) [1949], ICJ Rep 665, 701, para 81); James A Green, ‘Fluctuating Evidentiary Standards for Self-Defence in the International Court of Justice’ (2009) 58 ICLQ 168 (with respect to establishment of an armed attack); Oil Platforms (Islamic Republic of Iran v United States of America), Separate Opinion of Judge Higgins, ICJ Rep 2003, 233–34, para 32 (‘the Court did not even attempt to articulate the standard of proof it relied on”, merely referred to ‘insufficient’ evidence); Anna Riddell and Brendan Plant, Evidence before the International Court of Justice (BIICL 2009) 123–37. This ambiguity of the evidentiary standard is in keeping with the approach in civil law systems, which lack (rigid) evidentiary standards and in which what matters is the ‘intime conviction du juge’, Rosalyn Higgins, ‘The Judicial Determination of Relevant Facts’ in Rosalyn Higgins (ed), Themes and Theories, Selected Essays, Speeches, and Writings in International Law (OUP 2009) 1374; cf Juan José Quintana, Litigation at the International Court of Justice: Practice and Procedure (Brill Nijhoff 2015) 410.

66

Gabčíkovo–Nagymaros (n 41) para 54. The commentary on art 36(2) para 27, also uses ‘sufficient certainty’ in the context of the mitigating principle of ‘reasonable (or sufficient) certainty’.

67

Giancarlo Scalese, La rilevanza delle scusanti nella teoria dell’illecito internazionale (Editoriale Scientifica 2008) 121.

68

ARSIWA commentary art 25, para 16.

69

To use the words of the Court: Gabčíkovo–Nagymaros (n 41) para 54, which the ILC quoted ibid para 15.

70

Chester Brown, A Common Law for International Adjudication (OUP 2007) 101; Foster (n 50) 227.

71

ARSIWA commentary art 25, para 15.

72

ibid para 16.

73

See Ago, 1613th meeting, ILC YB 1980, vol I, para 6.

74

On which see Paddeu (n 12) 346–49.

75

Gabčíkovo–Nagymaros (n 41) para 53. For an overview, see Malgosia Fitzmaurice, ‘Necessity in International Environmental Law’ (2010) 41 NYIL 159.

76

ARSIWA commentary art 25, para 14.

77

Crawford (n 46) 307.

78

eg Enron (n 13) para 289; von Pezold (n 9) para 628.

79

See eg von Pezold (n 9) para 620.

80

eg Sempra (n 13) para 326; BG Group Plc v The Republic of Argentina, UNCITRAL, Award (24 December 2007) para 393; National Grid (n 13) para 245.

81

Von Pezold (n 9) para 631.

82

ibid para 8.49.

83

By way of example, see the following different approaches. Considering only the other State’s interests: Impregilo (n 61) para 354. Considering both the interests of the other co-contracting party and the investor: Enron (n 13) para 342; Sempra (n 13) para 392. Subsuming the investor’s interests as part of the other co-contracting party’s interests (ie the other BIT party has an interest in the protection of investments of its nationals): CMS (n 64) paras 357–58.

84

In at least one instance, the interest impaired was an interest of the international community as a whole: the interest in non-discrimination, von Pezold (n 9) para 657.

85

ARSIWA commentary art 12, para 3.

86

The German Federal Constitutional Court decided that no customary international rule entitles States to invoke necessity against individuals. According to the Court’s majority, necessity was limited to inter-State obligations, and did not extend to obligations between a State and individuals. The Court relied on the wording of ARSIWA Art 25, which notes that ‘[n]ecessity may not be invoked … as a ground for precluding the wrongfulness of an act not in conformity with an international obligation of that State’ (emphasis added). The Court reasoned that obligations against individuals—if they did not arise out of a bilateral treaty with another State—did not constitute international obligations of a State but were merely of a private character. Hence, the necessity defence did not apply to claims of private individuals. Decision of 8 May 2007, 2 BvM 1–5/03; NJW 2007.

87

Eran Sthoeger and Christian Tams, ‘Swords, Shields and Other Beasts: The Role of Countermeasures in Investment Arbitration’ (in this issue).

88

ARSIWA commentary art 25, para 17.

89

ibid para 1.

90

ILC, Report of the Commission on the Work of Its Thirty-Second Session, ILC YB 1980, vol II(2) 49 (para 33).

91

The final ARSIWA Commentary does not replicate this comment, but there is nothing in the Commission’s subsequent work to suggest that it subsequently rejected this view. The final Commentaries to ARSIWA provisions in Part One are often a simplified version of the commentaries of provisions adopted on first reading under Roberto Ago’s Special Rapporteurship. The ILC may have deleted the observation simply to streamline the commentary.

92

For an overview of the historical criticisms, see Paddeu (n 12) 374–77; Frz. Sales Eugen Steger, ‘Notwehr und Notstand im Völkerrecht’ (PhD, Universität Freiburg, 1927) 79–80; Klaus Radke, Der Staatsnotstand im modernen Friedensvölkerrecht (Nomos 1988) 49–52. The formulation was again controversial during the first reading of draft Art 33 (now Art 25). Since the adoption of ARS in 2001, see the criticism by Robert D Sloane, ‘On the Use and Abuse of Necessity in the Law of State Responsibility’ (2012) 106 AJIL 447, 478–81.

93

cf Caroline Foster, ‘Due Diligence and Compliance with the Protocol on Environmental Protection to the Antarctic Treaty’ (manuscript on file with the authors) 14.

94

For a similar critique in domestic law, see Alexander (n 10) 613 (fn 10).

95

This was not the only rationale discussed in the literature. Alongside so-called subjective rationales relying on fault, scholarship has put forward at least four other objective rationales for this defence, see Paddeu (n 12) 368–71.

96

Necessity is also available in private law: James Goudkamp, Tort Law Defences (Bloomsbury 2013).

97

eg German Criminal Code, §§ 34–35. The US Model Penal Code provision on necessity does not specify the relevant interest in its text, however the Commentary indicates that it regards death as a greater evil than bodily injury, bodily injury as greater than property damage; in addition, within these categories, the extent of harm differentiates evils: namely, two deaths are graver than one, and so on. See Model Penal Code, s 3.02, at cmts 1, 3.

98

Alexander (n 10) 614–15; Kenneth W Simons, ‘Exploring the Intricacies of the Lesser Evils Defense’ (2005) 24 Law & Philosophy 645, 646–49.

99

cf Simon Gardner, ‘Necessity’s Newest Inventions’ (1991) 11 Oxford J Leg Stud 125, 132–33. See for example the Canadian Supreme Court case of Perka v Queen, in which the Court chose to characterize necessity as an excuse as it did not think that courts had the legitimacy to decide the absolute or relative importance of conflicting interests, which was a matter for Parliament: Perka v The Queen (1984) 2 SCR 232. For a short comparative analysis between the international and the domestic criminal law defence of necessity—focused on common law jurisdictions—see Sloane (n 92) 476–78.

100

cf Valentin Jeutner, Irresolvable Norm Conflicts in International Law: The Concept of a Legal Dilemma (OUP 2017) 80 (necessity cannot be invoked when interests involved are equal or unknown).

101

Esmé Shirlow, Judging at the Interface: Deference to State Decision-Making Authority in International Adjudication (CUP 2021) 19–22 (legitimacy as an instrumental reason for deference).

102

For criticism, see von Pezold (n 9) paras 647ff.

103

ARSIWA commentary art 25, para 15.

104

ibid.

105

Crawford (n 77) 311.

106

ibid.

107

See further Section III.

108

Avidan Kent and Alexandra Harrington, ‘The Plea of Necessity under Customary International Law: A Critical Review in Light of the Argentine Cases’ in Chester Brown and Kate Miles (eds), Evolution in Investment Treaty Arbitration (CUP 2011) 254; Michael Waibel, ‘Two Worlds of Necessity in ICSID Arbitration: CMS and LG&E’ (2007) 20 LJIL 637, 638.

109

Enron Corporation and Ponderosa Assets LP v Argentine Republic, ICSID Case No ARB/01/3, Decision on Annulment (30 July 2010) paras 369–73.

110

This is different from alternative measures as part of primary law tests, eg in international trade law, Caroline Henckels, Proportionality and Deference in Investor-State Arbitration (CUP 2015) 57–62.

111

von Pezold (n 9).

112

For an excellent overview, see Federico Sturzenegger and Jeromin Zettelmeyer, Debt Defaults and Lessons from a Decade of Crises (MIT Press 2006) 165–86; Paul Blustein, And the Money Kept Rolling In (and Out): Wall Street, the IMF, and the Bankrupting of Argentina (PublicAffairs 2005) ch 9.

113

eg Suez Sociedad General de Aguas de Barcelona SA and InterAgua Servicios Integrales del Agua SA v Argentine Republic, ICSID Case No ARB/03/17, Decision on Liability (30 July 2010) para 238; Suez Sociedad General de Aguas de Barcelona SA and Vivendi Universal SA v Argentine Republic, ICSID Case No ARB/03/19, Decision on Liability (30 July 2010) para 260.

114

eg LG&E (n 63) para 257.

115

Urbaser SA and Consorcio de Aguas Bilbao Bizkaia, Bilbao Biskaia Ur Partzuergoa v Republic of Argentina, ICSID Case No ARB/07/26, Award (8 December 2016) para 716.

116

The Appellate Body has accepted that a WTO member may need to adopt a ‘suite’ of measures to address a policy problem. Regarding the necessity criterion under the general exceptions of Art XX GATT, see Appellate Body Report, Brazil—Measures Affecting Imports of Retreaded Tyres 172 (‘these measures already figure as elements of a comprehensive strategy designed by Brazil to deal with waste tyres. Substituting one element of this comprehensive policy for another would weaken the policy by reducing the synergies between its components, as well as its total effect’); subsequently referred to in Panel Report, Australia–Certain Measures Concerning Trademarks, Geographical Indications and Other Plainpackaging Requirements Applicable to Tobacco Products and Packaging 7.1384.

117

eg Enron (n 13) paras 308–09; Sempra (n 13) para 350.

118

See most clearly, Total SA v Argentine Republic, ICSID Case No ARB/04/1, Decision on Liability (27 December 2010), para 345; von Pezold (n 9) para 637.

119

See eg Enron (n 13) para 306 (the crisis was not grave enough because the government had adopted measures that had been effective in dealing with it).

120

Urbaser (n 115) para 717. Perhaps not coincidentally, in both cases the ‘only way’ criterion was met: von Pezold (n 9) paras 638–46; Unión Fenosa (n 7) para 8.46; Shirlow (n 101) 31–32, 170.

121

Continental Casualty (n 11) para 181.

122

Though note that more recent awards have paid more attention to this question: Unión Fenosa (n 7) paras 8.41–8.46; von Pezold (n 9) 638–46.

123

Waibel (n 108) 646.

124

Matthew Parish, ‘On Necessity’ (2010) 11 JWIT 169, 183.

125

Enron Annulment (n 109) para 370.

126

Though it may not always be good policy. See in the context of domestic law: Alexander (n 10) 619. For a different take, see Simons (n 98) 653–54.

127

See: John Gardner, ‘Fletcher on Offences and Defences’ Offeces and Defences: Selected Essays in the Philosophy of Criminal Law (OUP, 2007), 144–5.

128

Ryan Manton, Necessity in International Law (PhD thesis, University of Oxford, 2016, on file with the authors) 175.

129

ibid 107.

130

Indeed, this is the reason why the plea is excluded when the measure harms an essential interest of another State or of the international community as a whole: ARSIWA Art 25(1)(b).

131

Manton (n 128) 170–71.

132

Enron Annulment (n 109) para 371.

133

Tribunals can carry out this role. The Urbaser Tribunal (n 115) paras 724–32 thoroughly compared the alternative lawful measures suggested by Claimant.

134

Alan O Sykes, ‘Economic “Necessity” in International Law’ (2015) 109 AJIL 296, 299.

135

Gebhard Bücheler, Proportionality in Investor-State Arbitration (OUP 2015) 280 (referring to cases or State practice involving economic emergencies); Manton (n 128) 190.

136

In favour: Brazil: A/C.6/35/SR.47, para 20; Finland: A/C.6/35/SR.48, para 50; Italy: A/C.6/35/SR.49, paras 32–35. Against: Argentina: A/C.6/35/SR.51, para 3; FR Germany: A/C.6/35/SR.45, para 8; Israel: A/C.6/35/SR.50, para 14; Tunisia: A/C.6/35/SR.52, para 45.

137

See Enron (n 13) paras 311–12; Sempra (n 13) paras 353–54; National Grid (n 13) para 262; Impregilo (n 62) para 358; El Paso Energy International Company v Argentine Republic, ICSID Case No ARB/03/15, Award (31 October 2011) para 665; Suez and InterAgua (n 113) paras 241–42; Suez and Vivendi (n 112) paras 263–64.

138

CMS (n 64) para 329.

139

Enron Annulment (n 109) paras 385–93.

140

ARSIWA commentary art 25, para 20.

141

Explicitly Impregilo Spa v Argentine Republic (n 62) para 356. Implicitly CMS (n 64) paras 328–29; Enron (n 13) paras 311–12; Suez and InterAgua (n 112) paras 241–42; Suez and Vivendi (n 113) paras 263–64; El Paso (n 135) para 618.

142

Requiring at least negligence: LG&E (n 63) para 256; Enron (n 13) para 311; Sempra (n 13) para 353. Requiring intent or recklessness: Urbaser (n 115) para 711.

143

Requiring intent, see eg Roberto Ago, Eighth Report on State Responsibility—Add.5–7, ILC YB 1980, vol II(1) 20 (para 13); Pillitu (n 58) 315; Cassella (n 40) 168–69; requiring foresight of consequences: Manton (n 126) 201. Requiring some degree of negligence: Tarcisio Gazzini, ‘Foreign Investment and Measures Adopted on Grounds of Necessity: Toward a Common Understanding’ (2010) 7(1) TDM 19–22; Jorge E Viñuales, ‘Las cuestiones medioambientales y el concepto de estado de necesidad’ (2009) 119 Universitas 223, 235; Alberto Alvarez-Jiménez, ‘Foreign Investment Protection and Regulatory Failures as States’ Contribution to the State of Necessity under Customary International Law—A New Approach’ (2010) 27 J Int’l Arb 144, 159. Wilem Riphagen in the ILC’s debates on the predecessor to Art 25 seemingly took this view, see: Wilem Riphagen, 1614th meeting, ILC YB 1980, vol I, 161 (paras 6, 8).

144

The Commentary to draft Art 33, para 41 stated that ‘the Commission intended to refer to the case in which the State invoking the state of necessity has, in one way or another, intentionally or by negligence, contributed’ to the situation.

145

On which see Unión Fenosa (n 7) para 8.60.

146

Jorge E Viñuales, ‘Defence Arguments in Investment Arbitration’ (2020) 18 ICSID Rep 86.

147

Sloane (n 92) 481, 488. See also Waibel (n 107) 643 (in view of the impossibility of disentangling exogenous and endogenous causes of financial crises such as Argentina’s, the non-contribution requirement is ‘ill-suited to the economic field’).

148

Unión Fenosa (n 7) para 8.60.

149

See eg Maria Agius, ‘The Invocation of Necessity in International Law’ (2009) 56 NILR 95, 111; Kent and Harrington (n 108) 260; Manton (n 128) 201.

150

On which see eg Paolo Benvenuti, ‘Lo stato di necessità alla prova dei fatti’ in Marina Spinedi and others (eds), La codificazione della responsabilità internazionale degli Stati alla prova dei fatti (Giuffrè 2006) 144; Cassella (n 40) 164.

151

Roberto Ago, 1618th meeting, ILC YB 1980, vol I, 18, para 26.

152

Intentional contributions—namely, conduct adopted for the purpose of creating a situation of necessity—would exclude the plea, by application of the principle ex injuria, jus non oritur.

153

See eg Josef Kohler, Not kennt kein Gebot: Die Theorie des Notrechtes und die Ereignisse unserer Zeit (Rothschild 1915) 33–39 (defending Germany’s attack on Belgium as part of Germany’s ‘holy duty’ of self-preservation).

154

eg Steven R Weisman, ‘The Hazards of Moral Hazard’ in Steven R Weisman, The Great Tradeoff: Confronting Moral Conflicts in the Era of Globalization (Peterson Institute for International Economics, 2006); Tom Baker, ‘On the Genealogy of Moral Hazard’ (1996) 75 Tex L Rev 237; McCaffrey Matthew, ‘The Morals of Moral Hazard: A Contracts Approach’ (2017) 26 Business Ethics: A European Review 47; Benjamin Hale, ‘What’s So Moral about the Moral Hazard?’ (2009) 23 Public Affairs Quarterly 1; David Rowell and Luke B Connelly, ‘A History of the Term “Moral Hazard”’ (2012) 79 Journal of Risk and Insurance 1051.

155

The duty of compensation, as noted later, seems premised on the concept of moral hazard—at least as the ILC rationalized the duty in the ARSIWA commentary art 27, para 5.

156

ARSIWA commentary art 27, para 1. However, the facts which give rise to the defence may also be, at the same time, a sufficient basis for terminating the underlying obligation as a matter of treaty law: ibid, para 3.

157

CMS (n 64) 110, para 382. See also Gabčíkovo–Nagymaros (n 41) 7.

158

LG&E (n 63) paras 226–40.

159

Impregilo (n 62) paras 347–50.

160

Unión Fenosa (n 7) para 8.42.

161

ibid para 8.47.

162

LG&E (n 63) paras 221, 228, 263.

163

ibid para 261.

164

Hochtief AG v Argentine Republic, ICSID Case No ARB/07/31, Award (9 December 2014) para 295.

165

SAUR International SA v Argentine Republic, ICSID Case No ARB/04/4, Jurisdiction and Responsibility (6 June 2012), para 461.

166

Hochtief AG v Argentine Republic (n 164) para 294.

167

Stephen M Kissler and others, ‘Projecting the Transmission Dynamics of SARS-CoV-2 through the Postpandemic Period’ (2020) 368 Science 860.

168

Prakash S Bisen and Ruchika Raghuvanshi, Emerging Epidemics: Management and Control (Wiley 2013) 25.

169

Hochtief (n 164) para 293.

170

ibid para 294.

171

ibid.

172

See further Federica Paddeu and Michael Waibel, ‘The Final Act: Exploring the End of Pandemics’ (2020) 114 AJIL 698.

173

ARSIWA commentary art 27, para 2.

174

Insofar as justifications concern permissibility of conduct alone, as they do in international law, then they cannot be partial: conduct is either permissible or impermissible, by reference to the relevant rule and defence. Some domestic legal systems allow for the possibility of partial justifications: in these cases, the partial justification reduces the wrongfulness of the specific conduct (which remains a wrong). The effect is to reduce the gravity of the crime for which the defendant is convicted. Thus, a defendant who intentionally causes the death of the victim, and would normally be liable for murder, may be liable of a lesser offence of manslaughter if covered by a partial justification. This is possible because wrongfulness is often scalar within categories of offences: for example, the fatal offences against the person (homicide offences) are often arranged as a ladder, with the gravest (intentional killing) at the top and meriting the harshest penalties. Contrast this with the situation in international law, where there is no such scale, and where the question is usually whether the conduct is or is not in breach of a particular rule: if there is no lower rung in the ladder, the result of a partial justification would be that the conduct is only partially permitted, and therefore partially lawful. Such a conclusion would be illogical, and also cause some serious difficulties for the application of certain other institutions of international law (are, for example, countermeasures available against partially wrongful conduct?).

175

ARSIWA commentary art 12, para 2.

176

Note that the Commentary seems to limit the operation of this duty of compensation to inter-State relations: it thus refers to compensation due to ‘any State directly affected’: ARS Art 27, Commentary, para 4. This would potentially rule out any compensation to investors under this heading. Nevertheless, there is so much uncertainty in respect of the duty of compensation referred to in Art 27 that the question remains open.

177

ARSIWA commentary art 27, para 4.

178

Suggesting this legal basis Roberto Ago, 1612th meeting, 16 June 1980, ILC YB 1980, vol I, 154, para 46; Robert Q Quentin-Baxter, 1615th meeting, ILC YB 1980, vol I, 168 (para 16); Barboza, 1617th meeting, 176 (para 27). Against this view, Stephen M Schwebel, 1616th meeting, ILC YB 1980, vol I, 171 (para 12); Paul Reuter, 1614th meeting, ILC YB 1980, vol I, 163–64 (para 22).

179

ARSIWA commentary art 27, para 5.

180

CMS (n 64) para 390.

181

BG Group Plc v The Republic of Argentina (n 80) para 409.

182

EDF Award (n 13) paras 1171, 1177; EDF Annulment (n 13) para 330.

183

South American Silver (n 9) para 620.

184

LG&E (n 63) paras 106–08; CMS Annulment (n 11) paras 146–47.

185

Enron (n 13) para 345; Sempra (n 13) para 394.

186

Arguing that the duty has a customary basis: August Reinisch and Christina Binder, ‘Debts and State of Necessity’ in Juan Pablo Bohoslavsky and Jernej Letnar Černič (eds), Making Sovereign Financing and Human Rights Work (Hart 2014) 125–26; Bücheler (n 135) 243, 290–96; Christina Binder and Philipp Janig, ‘Investment Agreements and Financial Crises’ in Markus Krajewski and Rhea Tamara Hoffmann (eds), Research Handbook on Foreign Direct Investment (Edward Elgar 2019) 677–78. Arguing against: Florian Franke, ‘The Custom of Necessity in Investor-State Arbitrations’ in Rainer Hofman and Christian J Tams (eds), International Investment Law and General International Law (Nomos 2011) 156–57; Kent and Harrington (n 108) 261–63; Martins Paparinskis, ‘Investment Treaty Arbitration and the (New) Law of State Responsibility’ (2013) 24 EJIL 617, 633; Rodrigo Díaz Inverso, ‘El estado de necesidad como circunstancia que excluye la ilicitud en la responsabilidad internacional de los Estados’ (2015) 47 Revista de Derecho Público 49, 54; Paddeu (n 12) 421–25.

187

In addition to the sources in the previous two notes, see Sergey Ripinsky, ‘State of Necessity: Effect on Compensation’ (2007) 4(6) TDM 1; Jorge E Viñuales, Foreign Investment and the Environment in International Law (CUP 2012) 390.

188

For an overview, see Paddeu (n 12) 423–24.

189

Indus Waters Kishenganga Arbitration (Pakistan v India) PCA, Partial Award (18 February 2013) paras 397–99; Paddeu (n 12) 334–45. For example, in Marxist social theory necessity refers to the determination of outcomes by an external force such as the laws of physics, history or institutional arrangement, Roberto Mangabeira Unger, ‘Illusions of Necessity in the Economic Order’ (1978) American Economic Review 68(2) 369–73.

190

Oxford English Dictionary, ‘necessity, n.’.

191

Heathcote (n 4) 35–78.

192

Generally see Giuseppe Sperduti, ‘Introduzione allo studio delle funzioni della necessitá nel diritto internazionale’ (1943) 22 Riv DI 19; Heathcote (n 4) ch 4.

193

On the origins of these maxims, see Franck Roumy, ‘L’origine et la diffusion de l’adage canonique Necessitas non habet legem (VIIIe–XIIIe s.)’ in Wolfgang P Mueller and Mary E Sommar (eds), Medieval Church Law and the Origins of the Western Legal Tradition: A Tribute to Kenneth Pennington (Catholic University of America Press 2006) 457–86.

194

Many scholars have challenged the Schmittian notion that emergencies usher in extra-legality in the domestic sphere. See, most persuasively, David Dyzenhaus, The Constitution of Law: Legality in a Time of Emergency (CUP 2006).

195

Faber case (1903) 10 RIAA 438, 466.

196

Paddeu (n 12) vol 130, 335. See also Andrea Rapisardi-Mirabelli, I limiti d’obbligatorietà delle norme giuridiche internazionali (Giannotta 1922) 121–23; Massimiliano Montini, ‘La necessità ambientale e le sue diverse funzioni nel diritto internazionale contemporaneo’ in Marina Spinedi and others (eds), La codificazione della responsabilità internazionale degli Stati alla prova dei fatti (Giuffrè 2006) 159; Nicholas Tsagourias, ‘Necessity and the Use of Force: A Special Regime’ (2010) 41 NYIL 11, 12; Cassella (n 40) 461–62 (to avoid ambiguities, Cassella even rejects the use of sources language in this context); Sales Steger (n 92) 60.

197

In the sense used by Nguyen Quoc Dinh, Patrick Daillier and Allain Pellet, Droit international public (5th edn, LGDJ 1994) 111; International Law Association, Committee on the Formation of Customary (General) International Law, Final Report (2000) 12.

198

(1969) 970 UNTS 211.

199

See Heathcote (n 4) 172.

200

Jonathan Bonnitcha and others, The Political Economy of the Investment Treaty Regime (OUP 2017) table 4.1. Another major factor is likely the recognition by at least some States that first-generation investment treaties themselves, especially their treatment standards, safeguard insufficient regulatory space for States.

201

Levent Sabanogullari, General Exception Clauses in International Investment Law: The Recalibration of Investment Agreements via WTO-Based Flexibilities (Nomos 2018) 55 (a ‘modest trend’, out of 114 investment treaties with general exceptions concluded until the end of 2016, two-thirds (76) were concluded from 2000 onwards); Caroline Henckels, ‘Should Investment Treaties Contain Public Policy Exceptions?’ (2018) 59 B C L Rev 2826 (GATT XX-style exceptions feature in 43 per cent of agreements signed over 2011 to 2016, and only 7 per cent of agreements signed from 1959 to 2019).

202

These internal exceptions predetermine the weighing of the various interests in play, thus avoiding the ‘legitimacy problem’ of broad necessity rules, as explained in Subsection B (iii).

203

Wolfgang Alschner and Kun Hui, ‘Missing in Action: General Public Policy Exceptions in Investment Treaties’ in Lisa Sachs and others (eds), 2018 Yearbook on International Investment Law and Policy (OUP 2019) (noting that States fail to raise internal exceptions and tribunals downplay them). See Copper Mesa Mining Corporation v Republic of Ecuador, PCA No 2012–2, Award (15 March 2016); Bear Creek Mining Corporation v Republic of Peru, ICSID Case No ARB/14/21, Award (30 November 2017); CC/Devas (Mauritius) Ltd, Devas Employees Mauritius Private Limited and Telecom Devas Mauritius Limited v India, PCA Case No 2013–09, Award on Jurisdiction and Merits (25 July 2016); Deutsche Telekom v India, PCA Case No 2014–10, Interim Award (13 December 2017).

204

eg Georgia–Japan BIT (signed 29 January 2021, not yet in force) (‘Recognising that these objectives can be achieved without relaxing health, safety and environmental measures of general application’);

Hungary–Kyrgyzstan BIT (2020) (signed 29 September 2020, not yet in force).

205

Authors’ calculation, based on UNCTAD database of International Investment Agreements <https://investmentpolicy.unctad.org/international-investment-agreements>. The top 10 countries with the most investment treaties containing a public health exception are Germany (70), Canada (38), Singapore (22), Turkey (21), United Arab Emirates (21), India (20), Japan (20), Mauritius (15), Kuwait (13) and China (10). That said, there is considerable variation. The German variant of the exception is typically limited to the most-favoured nation treatment and national treatment. It is contained in a protocol which is an integral part of the investment treaty.

206

Federico M Lavopa and others, ‘How to Kill a Bit and Not Die Trying: Legal and Political Challenges of Denouncing or Renegotiating Bilateral Investment Treaties’ (2013) 16 JIEL 869; Timothy Meyer and Tae Jung Park, ‘Renegotiating International Investment Law’ (2018) 21 JIEL 655.

207

William W Burke-White and Andreas von Staden, ‘Investment Protection in Extraordinary Times: The Interpretation and Application of Non-Precluded Measures Provisions in Bilateral Investment Treaties’ (2008) 48 Va J Intl 343.

208

J Benton Heath, ‘The New National Security Challenge to the Economic Order’ (2020) 129 Yale LJ 1052.

209

Some authors limit the security exceptions to military threats, eg Prabhash Ranjan and Pushkar Anand, ‘Covid-19, India, and Investor-State Dispute Settlement (ISDS): Will India Be Able to Defend Its Public Health Measures?’ (2020) 28 Asia Pac L Rev 225 9. However, awards on necessity in Argentina’s financial crisis such as LG&E (n 62) provide some support to the view that there is no such limitation (‘the entire healthcare system teetered on the brink of collapse’, para 234) and Continental Casualty (n 11) para 180 (‘it is impossible to deny’ that ‘the immediate threats to the health of young children, the sick and the most vulnerable members of the population [amounted to] a situation where […] the protection of essential security interest [sic] as a state and as a country was vitally at stake’) suggest that security exceptions can cover non-military threats as well. LG&E (n 63) para 238 (‘When a State’s economic foundation is under siege, the severity of the problem can equal that of any military invasion’).

210

Burke-White and von Staden (n 207) 349.

211

ibid 361; cf also art 2(2) of the Agreement on Sanitary and Phytosanitary Measures (‘sanitary or phytosanitary measure [] applied only to the extent necessary to protect human, animal or plant life or health, [] based on scientific principles and [] not maintained without sufficient scientific evidence’). The WTO Appellate Body requires that SPS measures be based on scientific evidence or an appropriate risk assessment and a rational link between the evidence/assessment and the measures adopted.

212

Bonnitcha and others (n 200) 118.

213

eg India–Denmark BIT, art 12(2) (terminated) (‘necessary measures … in circumstances of extreme urgency for the prevention of diseases or pests’).

214

India–Mauritius BIT, art 11 (3) (terminated in 2017) (bundled together essential security interests).

215

ibid art 12.

216

Only time will tell. According to the UNCTAD database, only two investment treaties have been signed between March 2020 and January 2021.

217

Federica Paddeu and Freya Jephcott, ‘COVID-19 and Defences in the Law of State Responsibility Part II’ EJIL Talk (2020); Martins Paparinskis, ‘The Once and Future Law of State Responsibility’ (2020) 114 AJIL 618, 624–25.

218

Henckels (n 199) 2842; Ranjan and Anand (n 209) 11.

219

Devas v CC/Devas (Mauritius) Ltd, Devas Employees Mauritius Private Limited and Telecom Devas Mauritius Limited v India, PCA Case No 2013–09, Award on Jurisdiction and Merits (25 July 2016) para 354, based on art 11(3) India–Mauritius BIT, which also refers to ‘the protection of public health or the prevention of diseases in pests or animals or plants’.

220

Methanex Corporation v United States of America, UNCITRAL, Final Award (3 August 2005); Chemtura Corporation v Government of Canada, UNCITRAL, Award (2 August 2010).

221

eg 2004 US Model BIT, Annex B.4.b. (‘Except in rare circumstances, non-discriminatory regulatory actions by a Party that are designed and applied to protect legitimate public welfare objectives, such as public health, safety, and the environment, do not constitute indirect expropriations’); CETA, Annex 8-A.3; 2018 USMCA, Annex 14-B-3.b; India Model BIT, art 5.5; art 10, Argentina–Qatar BIT (‘[n]one of the provisions of [this] Agreement shall affect the inherent right of the Contracting Parties to regulate within their territories measures necessary to achieve legitimate policy objectives, such as the protection of public health’; Georgia–Japan BIT (signed 29 January 2021) (‘Non-discriminatory regulatory actions by a Contracting Party that are designed and applied to protect legitimate public welfare objectives, such as public health, safety and the environment, do not constitute expropriation, except in rare circumstances’). See also Julian Arato and others, ‘The Perils of Pandemic Exceptionalism’ (2020) 114 AJIL 627.

222

Henckels (n 201) 2835–37.

223

Indeed, until the mid-twentieth century, this practice addressed the principle reflected in the defence of necessity as part of the concept of force majeure. See Federica I Paddeu, ‘A Genealogy of Force Majeure in International Law’ (2012) 82 BYIL 381, 443–46.

224

Pinto, 1618th meeting—24 June 1980, ILC YB vol 1 (1980) 178, para 3. See also Constantin Economides, 2591st Meeting—22 June 1999, ILC YB vol 1 (1999) para 37 (‘The article was such a delicate balancing act that he did not see how it could be implemented in reality. So many things had to be proved that he thought it could perhaps be deleted’).

225

ILC, Report, 32nd session, 49, para 31.

226

Impregilo (n 62) paras 347–50.