Self-Judging Security Exception Clause as a Kind of Carte Blanche in Investment Treaties: Nature, Effect and Proper Standard of Review

Abstract

In investment treaties, the self-judging security exception clause allows states to restrict the exercise of investors’ rights and protections provided for by such treaties during security emergencies. During the last decade, such a clause has been included in numerous investment treaties to support state positions vis-à-vis foreign investors. States consider that this provision gives them a very broad discretion to limit or derogate from obligations which arise under the treaty. In this article, after having identified the self-judging character of this clause, it will be demonstrated that such a clause does not affect the jurisdiction of the tribunal but requires the tribunal to apply the principle of good faith as the proper standard of review to interpret the elements of the clause, preventing states’ abuse of such a clause.

The basic purpose of Bilateral Investment Treaties (BITs) is to protect foreign investors. However, states have always sought to preserve a range of their sovereign powers, especially in emergency situations, to balance the investor-state relationship. Accordingly, given the importance of security issues and their relationship to the very existence of a country, states tend to include the self-judging security exception clause in investment treaties.¹ Under this clause, states have the right of non-compliance with regard to international obligations and may take necessary measures in cases where their security interests are endangered. If such measures conflict with investors’ rights, this will not be considered a breach of treaty obligations. On the one hand, this clause allows the state to derogate from its international obligations under certain circumstances, while, on the other hand, the evaluation of such circumstances is not performed fully objectively from an external point of view, but primarily from the point of view of the state concerned.² Accordingly, the self-judging security exception clause can be defined as a provision in investment treaties under which a state has the right to unilaterally declare certain treaty obligations as non-binding if, according to its subjective determination, respecting the treaty obligations will harm its essential security interests. Thus, by means of the self-judging clauses, states are conferred with discretion to prioritize their security interests over international responsibilities.

Before addressing categories of the security exception clause, it is worth articulating the structure of such clauses. Even though there are slight differences in the formulations of the security exception clauses in the BITs of various countries, they contain similar conditions that have to be met. According to some scholars, security exception clauses generally contain two elements: (1) a permissible objective and (2) a nexus requirement.³ As far as the first element is concerned, a permissible objective is considered the ultimate goal for the adoption of measures contrary to a state's treaty obligations. In the security exception clause, as the title indicates, security interests are considered permissible objectives. Regarding the second element, the nexus requirement mandates that measures taken by a state must be appropriately related to security interests. This denotes that there must be a link between the measures taken and the security interests’ objectives.

As to the basis of how these two elements are drafted in relevant treaty provisions, two main self-judging security exception clauses can be recognized. First, some states, particularly the United States (US), have bluntly grafted World Trade Organization (WTO) type exceptions onto their International Investment Agreements (IIAs) through the reproduction of texts from Articles XXI of the General Agreement on Tariffs and Trade (GATT). Under these GATT-type clauses, the nexus requirement is self-judging in nature. It signifies that it is the state that determines which measures are necessary: arbitration tribunals cannot substitute their judgment for that of the state. Such clauses are characterized as partly self-judging clauses because the assessment of the first element, i.e. security interests, still lie within the tribunal's discretion. For instance, Article 1.5 of the European Union-Japan Economic Partnership Agreement provides that “[n]othing in this Agreement shall be construed … as preventing a Party from taking any action which it considers necessary for the protection of its essential security interests…”⁴ (emphasis added). As can be seen, the contracting parties identified the self-judging character for the nexus requirement by stating that a state can take actions “it considers necessary” to protect its security interests. Therefore, the authority for determining the necessity of measures would be the state invoking the security exception clause.

Second, the United States-Korea Free Trade Agreement (KORUS) clauses, which are fully self-judging, imply that the state determines not only the necessity of measures, but also what constitutes security interests as well as other elements contained in the clause. Such clauses can exclusively be found in the US treaty practice. In this regard, KORUS provides that:

Nothing in this Agreement shall be construed: to preclude a Party from applying measures that it considers necessary for the fulfillment of its obligations with respect to the maintenance or restoration of international peace or security or the protection of its own essential security interests.

This provision reads together with its footnote, which states:⁵

For greater certainty, if a Party invokes Article 23.2 in an arbitral proceeding initiated under Chapter Eleven (Investment) or Chapter Twenty-Two (Institutional Provisions and Dispute Settlement), the tribunal or panel hearing the matter shall find that the exception applies.

Similar provisions also appear in the United States-Colombia Trade Promotion Agreement,⁶ the United States-Peru Trade Promotion Agreement (TPA),⁷ and the United States-Panama TPA.⁸ By its terms, the footnote appears to extend the scope of a state's discretion to all elements of the security exception clause and correspondingly limit the review of arbitral tribunals.

The self-judging security exception clause in new investment treaties could confront tribunals with some types of interpretative challenges. First, it is unclear whether these clauses affect the tribunal's jurisdiction or standard of review by lowering it from a substantive analysis to a good faith analysis. Particularly, one may ask whether the footnote of a KORUS-type clause would render a dispute over security matters non-justiciable. Second, in the case of negative response, it is unclear how tribunals should interpret elements of these clauses and concretize relevant standards of review since the clause limits the discretion of the tribunal, preferring the subjective determination of the state.

At first glance, it seems that such clauses give states a broad discretion so that they are able to easily escape from international obligations, but a scrutinized examination reveals the important role of international tribunals in adopting a proper view on the interpretation of such provisions. This article thus challenges the prevailing view in state practice. While states are to be given a margin of discretion as to what constitutes a threat to their security interests and what kind of measures are necessary to protect those interests, adjudicators are tasked with the job of controlling the exercise of such a discretion. In this regard, this discretion does not primarily mean that a tribunal has no jurisdiction over the dispute. Moreover, it ought not to allow states to invoke the self-judging security exception provision in times of socio-economic emergencies or where there is no rational link between the challenged measure and the end pursued. The article argues how tribunals should regard these clauses to avoid any abuse of them. To this end, it proposes conceptual and textual legal tools to control states’ discretion in a justified manner.

The article primarily focuses on the consideration of the security exception clause in investment arbitrations, but also considers the extent to which the approach taken by other international tribunals on this issue may be helpful in resolving the issues faced within international investment law. In fact, arbitral tribunals can be duly conceived as creatures of public international law and, as Crawford highlights, “[i]nvestment law … is about the way in which we bring the state under some measure of control, which is the main aspiration of general international law”.⁹ Accordingly, we believe that arbitral tribunals can consider the system of international law and especially the available body of WTO law to specify the appropriate standard of review. There are several reasons to support this assertion. First, “international investment law and WTO law belong to the same branch of international law, namely international economic law”.¹⁰ Second, the nature of the issues facing both systems is similar. Investment arbitration tribunals as well as WTO panels and their appellate body often assess states’ actions that were adopted in support of their essential interests against a set of obligations of a purely economic character. Third, WTO jurisprudence on the standard of review applicable to the same provisions is well established.¹¹ Finally, and most importantly, states incorporate the text of Article XXI of the GATT into their investment treaties, particularly as regards WTO-type clauses. WTO case law on the interpretation and application of the security exception in Article XXI of the GATT is arguably relevant to the interpretation of such exceptions within IIAs.¹² Where an IIA reproduces the WTO equivalents verbatim, WTO case law could conceivably be relevant in determining the “ordinary meaning” of terms within the WTO security exceptions pursuant to Article 31(1) of the Vienna Convention on the Law of Treaties (VCLT). WTO case law might also be regarded as relevant to the interpretation of security exceptions in IIAs, either as a subsidiary means of identifying the relevant rules of international law applicable between the IIA parties pursuant to Article 31(3)(c) of the VCLT and Article 38(1)(d) of the International Court of Justice (ICJ) Statute, or as a supplementary means of interpretation pursuant to Article 32 of the VCLT.¹³ On this basis, in this article we consider it both necessary and appropriate to examine not only investment jurisprudence but also decisions of the ICJ and WTO dispute settlement bodies.

The remainder of this article is structured as follows: initially, it attempts to articulate how tribunals determine the self-judging character of a treaty provision. It then demonstrates the legal effect of such a clause on the tribunal's jurisdiction or standard of review in the merits phase. Finally, by clarifying elements of the clause, the article concludes by proposing proper interpretations applicable to each of these elements.

I. Determining the Self-Judging Character

To identify the self-judging nature of a treaty clause it is necessary to consider the rules of treaty interpretation in Articles 31 and 32 of the VCLT. In investment arbitration proceedings and treaty practice, two ways of introducing the self-judging element can be identified. First, states may employ explicit wording such as “it considers necessary” or “it determines necessary” to indicate its self-judging nature. Second, as an implicit way, the US practice was invoked by Argentina to establish such a character for Article XI of the US-Argentina BIT.

A. Explicit Self-Judging Clause

First, according to paragraph 1 of Article 31, each treaty shall be interpreted in accordance with the ordinary meaning to be given to the terms of the treaty in their context and in the light of its object and purpose. Considering this rule, in investment treaties as well as in other international treaties, states emphasize the self-judging character by employing the explicit wording “the state considers necessary”.¹⁴ Accordingly, in US investment treaties before 1992, the words “measures necessary for” were frequently used,¹⁵ but the US changed this practice from 1992 onwards after the ICJ in the Nicaragua case held that such a clause was non-self-judging.¹⁶ For the first time, in the protocol to the BIT with Russia, the parties explicitly recognized the security exception clause as a self-judging provision.¹⁷ However, the phrase “the state considers necessary” was first used in the original text of the BIT with Mozambique¹⁸ and, from this date on, this formulation is taken to express the self-judging nature of security exception clause in most US BITs.¹⁹ Finally, in the 2004 and 2012 US Model BITs, the phrase “the state considers necessary” has been included.²⁰

In other investment treaties, especially those of the European Union (EU), Japan, the UK, and Iran, the contracting parties have also chosen this formulation to express their intention to establish a self-judging character for the security exception clause.²¹

This practice is not only observed in investment treaties, but also in other areas of international law. In this regard, in the field of trade, Article XXI of the GATT, using identical wording, empowers members to take measures which they consider necessary for their essential security interests.²² Similarly, Article 2 of the Mutual Assistance in Criminal Matters between France and Djibouti sets out that “[the request] may be refused … if the requested State considers that the execution of the request is likely to prejudice its sovereignty, its security, its order public or other of its essential interests”.²³ Besides, Article 346 of the Treaty on the Functioning of the European Union (TFEU) 1957 provides that “any Member State may take such measures as it considers necessary for the protection of its essential interests”.²⁴

In the Nicaragua case, the ICJ also upheld the importance of the wording used in the treaty to identify the self-judging nature of the particular clause. The US-Nicaragua Friendship, Commerce, and Navigation (FCN) treaty included a provision that the parties to the treaty could take the necessary measures for their security interests, which the US invoked to contest the Court's jurisdiction in dealing with security issues. In that case, the ICJ, by emphasizing the terms used in this article and considering the lack of the phrase “the state considers necessary”, ruled that:²⁵

This provision of GATT (Article XXI) … stipulates that the Agreement is not to be construed to prevent any contracting party from taking any action which it ‘considers necessary’ for the protection of its essential security interests, in such fields as nuclear fission, arms, etc. The 1956 treaty, on the contrary, speaks simply of ‘necessary measures’, not of those considered by a party to be.

It can be concluded that, in this judgment, one of the ways to identify a self-judging character is the explicitness of the terms of the treaty. In a similar vein, the Court acknowledged its view in the Mutual Assistance case by stating that the terms of Article 2 provide a state, to which a request for assistance has been made, with a very considerable discretion.²⁶

As regards fully self-judging clauses, the US typically uses a particular footnote to indicate that all elements of the security exception clause have a self-judging character. For instance, the footnote of the KORUS treaty clarifies that “[f]or greater certainty, if a Party invokes Article 23.2 in an arbitral proceeding … the tribunal or panel hearing the matter shall find that the exception applies”. This footnote confirms the intention of parties to limit the arbitral review concerning all elements of the provision.

Investment arbitration tribunals in disputes related to the financial crisis of Argentina, referring to the Nicaragua judgment, considered the express wording as the decisive element to identify a self-judging clause. Article 11 of the Treaty provided that the parties can take the necessary measures to protect their security interests and public order.²⁷ The CMS tribunal, in taking Article XXI of the GATT as an eloquent example, stated that “when states intend to create for themselves a right to determine unilaterally the legitimacy of extraordinary measures importing non-compliance with obligations assumed in a treaty, they do so expressly”.²⁸ Similarly, in the Enron and Sempra cases, the arbitral tribunals remarked that truly exceptional and extraordinary clauses, such as a self-judging provision, normally must be expressly drafted in the treaty so as to reflect the intent of the parties.²⁹ In the Deutsche Telekom case, the arbitral tribunal, in interpreting a similar clause found in the India-Germany BIT, stressed the absence of the phrase “the state considers necessary” in the treaty and concluded that “clear indications of the treaty would be required in order to infer that a provision is self-judging”.³⁰

From a general point of view, the requirement of unequivocal wording indicating the intention to recognize a self-judging character for a treaty clause emanates from a textual treaty interpretation approach.³¹ The approach advocates that the determination of a treaty clause's meaning must follow from the text and the actual words used by the parties. Article 31(1) of the VCLT endorses a textual approach, which gives prevalence to the treaty text over “the external will” of the parties.³² States are assumed to know what they are employing in their treaty and to be capable of expressing their intentions in clear terms.³³ Moreover, as a treaty elucidates the common will of the contracting parties, its meaning cannot be explored in unilateral determinations by only one of the parties.³⁴ As the tribunal stated in the Mobil case:³⁵

… the wording of the treaty is deemed to express the intention common to the Parties, and what the Parties effectively agreed to, even though a Party may have wished something else on one or another point. This means that, unless the contrary is specified, they are not self-judging.

B. Implicit Self-Judging Clause: Revisiting the Argentine Cases

In the Argentine cases it was alleged that another way to establish a self-judging character for a treaty clause is to consider the subsequent practice and actions of the state in question, in accordance with paragraph 3 of Articles 31 and Article 32 of the VCLT. In fact, the question to be addressed was to whether the incorporation of the phrase “it considers necessary” in newer US treaties and other treaty documents should have a bearing on the interpretation of previous agreements such as the US-Argentina BIT of 1991.

The application of this method, especially in the case of US practice, has been controversial in arbitrations and doctrine. The US, in FCN treaties and its BIT models of 1984, 1987, 1992, and 1994, used the phrase “necessary measures for” in its security exception provisions which, based on the previous discussion, is not self-judging. However, in the Nicaragua case, the US argued before the ICJ that the security exception clause contained in the FCN treaty between the two countries, despite not being specified, was of a self-judging nature.³⁶ As previously noted, the Court rejected this argument. That was why, two months after that judgment, the State Department spokesman told the Senate Foreign Relations Committee that the US considered the essential security interests provision as a self-judging clause.³⁷ The Reagan Administration confirmed this view.³⁸ Also, ever since the US-Russia BIT, concluded in 1992, the US has explicitly included the phrase “the state considers necessary” in its treaties and has applied this change to the 2004 and 2012 Model BITs.³⁹ Likewise, the US – instead of using the text of the treaty – has stated in its letters of submittal accompanying US BITs, forwarded to the Senate for ratification, that the provision at issue was characterized as self-judging.⁴⁰

The State Department also issued a formal statement interpreting each of the terms of the BIT model, which it introduced in 1992. As previously noted, in the 1992 Model BIT, the US had not yet used clear and explicit terms for recognizing the self-judging nature of a treaty, but explained in its statement that “[w]e are careful to note, in each negotiation, the self-judging nature of the protection of a Party's essential security interest”.⁴¹

The question concerning US practice is whether these actions should be considered as the subsequent practice or supplementary means in the interpretation of a treaty provision, and whether this will change the nature of the treaty clause. Some scholars have proposed the theory of an implicit self-judging clause based on US practice.⁴² According to this view, in the case of treaties such as the US-Argentina BIT, the US practice from 1992 onwards implicitly endorses the self-judging character for the security exception clause.⁴³ Argentina, in the International Centre for Settlement of Investment Disputes (ICSID) arbitrations, relying on this theory as well as the dissenting opinions of Judges Huber and Anzilotti, contended that the security exception clause is self-judging.⁴⁴ In the Wimbledon case, Judges Huber and Anzilotti issued a joint dissent stating that “[t]he right of a state to adopt the course which it considers best suited to the exigencies of its security and to the maintenance of its integrity, is so essential a right that, in case of doubt, treaty stipulations cannot be interpreted as limiting it”.⁴⁵

Contrary to this view, Alvarez argued that nothing in the text and negotiation history of the treaty implies a self-judging clause, and that the said records are set out in other circumstances which have nothing to do with this treaty.⁴⁶ Furthermore, the US-Argentina BIT was concluded based on the 1987 treaty model and, to interpret it, one cannot refer to the State Department statement on the 1992 treaty model.⁴⁷ Vandevelde characterized the US position in the Nicaragua case as a temporary shift in US policy prompted by the highly-charged atmosphere of the Nicaragua case to defeat jurisdiction in that case.⁴⁸ He argued that there was no public record that any US BIT partner ever had been informed that the US regarded the essential security interests exception as self-judging.⁴⁹ He concluded that if we seek to give a special meaning to the provisions of the treaty according to the rules of treaty interpretation, we need conclusive and decisive proof on the present facts. It is evident that the total public silence on the part of the US on this issue after 1986 does not constitute the decisive proof needed to establish a specialized meaning.⁵⁰

To ascertain whether US actions form subsequent practice within the meaning of paragraph 3 of Article 31 of the VCLT, it must first be determined whether these unilateral actions can be considered as subsequent practice and, second, whether they are related to the US-Argentina BIT. The ICSID tribunals in the Argentine cases responded negatively to both questions. In this regard, the Mobil tribunal held that “changes occurring elsewhere, for instance negotiations concerning other treaties, are irrelevant for establishing the meaning given to the provision”.⁵¹ Similarly, in the El Paso case the tribunal stated that the US records “did not relate to the interpretation of the 1991 BIT but to the American attitude in future negotiations with other states”.⁵² Moreover, as held by the Sempra tribunal, the view of one state does not make international law,⁵³ and in this regard, the El Paso tribunal clarified that “[i]nternal changes between the organs of one Contracting State, coupled with the silence of the other State, are not sufficient to produce an agreed practice of interpretation”.⁵⁴

Additionally, the supplementary means of interpretation may be used to establish a special meaning, or to confirm or invalidate the interpretation obtained by applying Article 31, or to correct results so obtained if they are ambiguous or obscure. First, by applying the elements listed in Article 31 of the VCLT and, given the ICJ judgment in the Nicaragua case, the meaning of the language “measures necessary for” is not ambiguous, or obscure, or wrong. Second, the US records could not be considered as “conclusive or decisive proof” to establish such a special meaning for the security exception clause as they were travaux preparatoires, or circumstances surrounding the conclusion of other treaties, but not with Argentina. On the contrary, the circumstances at the time of concluding the US-Argentina BIT indicate the non-self-judging character of Article XI because, at that time, the ICJ ruled that such clauses were not self-judging and the parties being aware of this fact negotiated and concluded the BIT. In this regard, the El Paso tribunal when reviewing this issue found that:⁵⁵

…no travaux préparatoires … have been identified—declarations made in connection with other BITs or the 1992 Model Treaty do not qualify as preparatory work related to the present Treaty—and the same is true for the circumstances surrounding the latter's conclusion.

In the Mobil case, the tribunal did not find that the interpretation based on Article 31(1) to (3) of the VCLT was wrong, obscure, or ambiguous to justify recourse to the supplementary means.⁵⁶ Besides, it held that:⁵⁷

[n]o preparatory work to the contrary has been identified. Declarations made in connection with other BITs or the 1992 Model Treaty do not qualify as preparatory work related to the present BIT. Neither have any circumstances surrounding the present BITs conclusion been shown that would contradict the same conclusion.

Therefore, it can be inferred from these arbitral awards that in the case of the US-Argentina BIT and similar BITs that the US shift in ideology did not establish subsequent practice, and the records are not considered as supplementary means of interpretation. Another issue that arises is that if, for example, the subsequent practice is formed within the meaning of Article 31, can the relevant clause be considered as self-judging? The answer to this question will be negative in the case of established investments.⁵⁸ In fact, the investor is considered to be a third-party beneficiary in relation to the investment treaty, and the subsequent practice of the parties in recognizing the nature of self-judging should not affect the acquired rights of the investor.⁵⁹ As held in the El Paso case, BITs are about funds invested by foreign nationals, not about direct relationships between states; therefore, self-judging exceptions such as that allegedly introduced into Article XI of the Argentina-US BIT by tacit consent would give extremely large powers to the state in whose territory the investment was made – the reference to good faith is of little help – and expose investors to large risks.⁶⁰

II. Effect on Jurisdiction or on Standard of Review

In this section, we address the question of whether the self-judging security exception clause will affect the tribunal's jurisdiction or the standard of review by lowering it from the level of substantive to good faith analysis. It is argued that such clauses are excluded from the jurisdiction of the dispute settlement body that adjudicates them. The most important arguments for this claim fall into three categories. First, as the security exception clause is permissive in nature, the obligations set out in the treaty do not apply where a state establishes that its actions come within the scope of the security exception clause.⁶¹ Therefore, it can be said that the applicability of the clause ought to be considered as a preliminary question of jurisdiction. The second is based on the explicit wording used in such a clause and on its self-judging nature. The existence of the phrase “it considers necessary” or other similar wording suggests that it is the state that determines its own security interests, and what measures are necessary to protect those interests. Most importantly, it could be argued that the footnote of the KORUS-type clause deprives a tribunal of jurisdiction over a dispute, and it operates as a limitation ratione voluntatis. We will discuss them in the following subsections.

A. Security Exception Clause as a Limitation on the Scope of the Treaty Obligations

Given the terms used in a security exception clause, the clause may be characterized as an internal limitation on the scope of substantive obligations within a treaty. In other words, if state measures are covered by this clause, treaty obligations do not apply.⁶² Accordingly, it is argued that the tribunal should interpret this clause in an investment treaty as a “jurisdictional threshold question that determines the scope of an arbitral tribunal's competence”.⁶³ The argument seems to be that a security exception clause limits the jurisdiction of the tribunal ratione materiae.

The ICJ had to assess the relevance of this argument on four occasions. In these cases, the exception was contained in an FCN treaty concluded by the US with Nicaragua and Iran, respectively. The language of the security exception clause in these two FCN-treaties is identical: “[t]he present Treaty shall not preclude the application of measures: […] (d) necessary to fulfil the obligations of a Party for the maintenance or restoration of international peace and security, or necessary to protect its essential security interests”.⁶⁴

In all four cases, the US refused to recognize the jurisdiction of the ICJ because of the existence of this clause.⁶⁵ However, the ICJ held that the invocation of the clause did not deprive the court of jurisdiction which, instead, is a matter for examination at the merits phase of the case. The ICJ has rejected the US argument in all four cases and ruled in the Nicaragua case that “being itself an article of the Treaty, it is covered by the provision in Article XXIV [dispute settlement provision] that any dispute about the ‘interpretation or application’ of the Treaty lies within the Court's jurisdiction”.⁶⁶ The ICJ clarified its view in the Oil Platforms case and held that Article XX did not restrict its jurisdiction, but confined it to affording the parties a possible defence on the merits to be used should the occasion arise.⁶⁷ In the Alleged Violations of the Treaty of Amity case, the Court stated that:⁶⁸

…whether and to what extent those exceptions have lawfully been relied on by the Respondent in the present case is a matter which is subject to judicial examination and, hence, forms an integral part of the material scope of the Court's jurisdiction.

As can be seen, the ICJ considered the national security exception as a part of the treaty, which is subject to the dispute settlement clause, and thus has jurisdiction over it. Moreover, it has made it clear that the security exception clause can be invoked as a defence in the merit stage. In the Argentine financial crisis cases and two Indian cases related to the Devas Project, the arbitral tribunals have found themselves competent to examine such clauses.

B. Effect of the Adjectival Clause: “It Considers Necessary”

The second argument is based on the explicit wording of “it considers necessary”, which is used in such a clause. States have contended that the existence of the language “it considers necessary”, or other similar wording, suggests that it is the states that determine their security interests and what measures are necessary to protect those interests.

In this regard, the question of how the Court interprets this adjectival clause was addressed in the Mutual Assistance case when the ICJ introduced an important clarification to its jurisprudence with respect to the effect of the language used in self-judging clauses. In that case, the ICJ was tasked with examining whether France's denial of the legal assistance requested by Djibouti was in breach of the mutual assistance conventions between the two countries. Relying on Article 2(c) of the Mutual Assistance Convention, the French courts denied the request by noting that it was considered to be contrary to the essential interests of France in that the file contained declassified defense secret documents, together with information and witness statements in respect of another case in progress.⁶⁹ Even though the ICJ accepted that Article 2(c) conferred a wide discretion on a state to refuse mutual assistance, it held that France's exercise of that discretion remained subject to a good faith review, as established in Article 26 of the VCLT.⁷⁰ This decision shows that the partly self-judging clause only affects the standard of review applied by the ICJ in the merits phase and that, instead of adopting a narrow standard, it should simply determine whether the principle of good faith has been respected. As a result, this clause does not preclude the Court's jurisdiction.

In GATT/WTO jurisprudence,⁷¹ despite continuing disagreement between members, panels considered themselves competent to deal with disputes.⁷² In the dispute between Ukraine and Russia concerning the interpretation of Article XXI, the panel considered that “Article XXI(b)(iii) of the GATT 1994 is not totally self-judging” since, in the view of the panel:⁷³

the adjectival clause ‘which it considers’ in the chapeau of Article XXI(b) does not extend to the determination of the circumstances in each subparagraph (objective criteria). Rather, it qualified the first element of the article, namely the necessity of actions.

In this report, the panel's jurisdiction was confirmed and, in its subsequent analysis, it found that the language “it considers necessary” affects the standard of review, which then only requires a determination on whether the decision was arrived at in good faith.⁷⁴ In a recent case, which involved a dispute between Saudi Arabia and Qatar concerning violation of intellectual property rights, Saudi Arabia invoked the security exception in Article73(b)(iii) of the Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS Agreement), which is identical to Article XXI of the GATT. The panel took the same approach and found good faith analysis to be an appropriate standard of review.⁷⁵

The jurisprudence of investor-state arbitration tribunals also underlines the view that a partly self-judging clause does not constitute a bar to the jurisdiction of international dispute settlement bodies but, rather, affects the standard of review by lowering it from substantive to good faith analysis. For instance, the tribunals in the Argentine cases expressed that, even though they had been tasked with assessing Argentina's measures under a self-judging clause of the US-Argentina BIT, they would have had the power to review them for good faith.⁷⁶

C. Effect of the KORUS-Type Footnote on Jurisdiction

Given the fact that KORUS-type clauses are fully self-judging in nature, according to which the tribunal must find that the exception applies, one could argue that such clauses operate as a limitation on jurisdiction ratione voluntatis. We reject this contention for several reasons. First, as some scholars have affirmed, “such clauses require the host-State to ‘invoke’ the clause. Such an invocation, like any other exercise of right under a treaty would have to be in good faith.”⁷⁷ In fact, the footnote stipulates that “if a party invokes Article …”, meaning that it does not render the clause an automatic defense and the tribunal does not have to consider the applicability of the article on its own initiative. If a KORUS-type clause is a jurisdictional threshold matter it is up to the tribunal to raise it to confirm its own jurisdiction.

Second, the US did not seek to consider such clauses as a jurisdictional limitation. The US Congressional Research Service, in its study on Dispute Settlement provisions of the KORUS treaty, remarked that the security exception clause is “self-judging in nature, although each government would expect the provision to be applied by the other in good faith”.⁷⁸ As Vandevelde highlighted, the US itself has, in the past, taken the position that self-judging reservations are subject to a requirement of good faith and the record precludes any argument that a self-judging provision cannot be made subject to an obligation of good faith.⁷⁹ Considering the term “for greater certainty”, it can argue that such a footnote serves as a clarification of the intention of the parties, not as an additional element to indicate a new shift in US BIT ideology. It should be noted that the US employs the same wording to clarifying different treaty provisions in its Free Trade Agreement and Trans-Pacific Agreement treaties.

Third, based on a United Nations Conference on a Trade and Development (UNCTAD) study, if the parties sought to exclude judicial review of the invocation of the national security exception, they should employ explicit wording to that effect.⁸⁰ Some treaties have followed such a practice. One good example is the one that UNCTAD mentioned is the Comprehensive Economic Cooperation Agreement between the Republic of India and the Republic of Singapore (2005). The Agreement states that “this Article shall be interpreted in accordance with the understanding of the Parties on non-justiciability security exceptions as set out in their exchange of letters, which shall form an integral part of this Agreement”.⁸¹

Fourth, in the Norwegian Loans case, Judge Lauterpacht argued in his separate opinion that:⁸²

an instrument in which a party is entitled to determine the existence of its obligation is not a valid and enforceable legal instrument of which a court of law can take cognizance. It is not a legal instrument. It is a declaration of a political principle and purpose.

This view, which is also enshrined in the law of obligations of most legal systems, can also be properly applied to the security exception clause. According to this theory, if a BIT-party can excuse itself from BIT obligations at will, then it can be argued that the entire treaty is illusory; that is, the parties in fact have bound themselves to nothing more than a meaningless requirement that they go through, a charade citing national security.

Accordingly, it can be argued that a KORUS-type clause does not deprive a tribunal of jurisdiction over a dispute and that a state's decision pursuant to a KORUS-type clause is still subject to good faith. One crucial question is whether the footnote changes the scope of the good faith analysis. In other words, does it permit an inference that the parties to a fully self-judging provision may have intended that the footnote excludes such clauses from the obligation of good faith? The answer is negative for two main reasons. First, it could be argued that the obligation to perform treaty provisions in good faith, which stems from the principle of pacta sunt servanda, is a peremptory norm of international law. As such, the obligation of good faith cannot be modified by the agreement of the parties.⁸³ Second, even if the obligation of good faith is not a peremptory norm, simple logic dictates that, as a concept implicit in the principle of pacta sunt servanda, the good faith obligation cannot be disavowed by the agreement of the parties. A treaty that purports to exempt itself from the obligation of good faith, and hence the principle of pacta sunt servanda, would represent a logical conundrum because its validity would appear to depend on the very norm which it purports to abolish.⁸⁴

The conclusion to be drawn from the international decisions is that the self-judging clause of national security cannot oust the jurisdiction of investment arbitration tribunals, but it does affect the standard of review applied by tribunals in the merits stage.

III. Proper Standard of Review: Interpretation Framework of the Elements of the Self-Judging Security Exception Clause

At first glance, it seems that the self-judging clause of national security provides states with a kind of carte blanche – that is, a broad discretion to prioritize their domestic interests over international responsibilities. By identifying and distinguishing the elements of this clause, however, it will become clear that this carte includes some restrictions that need to be applied. In this section, we will discuss the appropriate standard of review for addressing the elements of the self-judging security exception, namely security interests and the nexus requirement.

A. Permissible Objective: Essential Security Interests

As argued above, all the elements of a KORUS-type clause have a self-judging character. As such, it is the state that determines what kind of matters are related to its security interests. However, the footnote does not allow any interest to become an essential security interest, but requires a good faith basis for such a classification. Concerning the partly self-judging clause, the first issue to be considered by tribunals is how the essential security interests in a standard security exception provision⁸⁵ are defined and what situations they encompass. The following analysis will demonstrate that tribunals have recognized a broad scope for security interests and treated this element in a partly self-judging clause as identical to that of a KORUS-type clause. This section analyzes how international tribunals have interpreted this element in their jurisprudence, and then explains the authors’ argument on the factors that tribunals should take into account in the application of the relevant standard of review, such as good faith, to avoid the abuse of security exception clauses by contracting parties.

In the Nicaragua case, the ICJ, when interpreting this element contained in Article 11 of the FCN treaty, stated that “the concept of essential security interests certainly extends beyond the concept of an armed attack, and has been subject to very broad interpretations in the past”.⁸⁶ In a similar vein, in the Oil Platforms case the Court found that the US considered the security interests to include “the safety of United States vessels and crew, and the uninterrupted flow of maritime commerce in the Persian Gulf” and that Iran recognized these as being reasonable security interests.⁸⁷ It can be inferred from this judgment that the Court even considered “an uninterrupted flow of maritime commerce” to be related to the national security of states and has not limited the concept of national security to territorial integrity and defense against military invasions.

The approach of the ICJ has also been confirmed in ICSID arbitrations. In the Argentine cases American investors believed that the economic crisis did not fall within the concept of “essential security interests”; instead, it only included war, natural disasters, and other issues that endangered the existence of the state.⁸⁸ In contrast, Argentina argued that the concept should be interpreted broadly to include situations such as economic crises.⁸⁹ The arbitral tribunals dealing with these cases adopted Argentina's arguments, stating that a state of emergency could be of an economic nature.⁹⁰ In the Continental award, the tribunal, relying on the gravity of the crisis and its consequences on the country, concluded that:⁹¹

it is well known that the concept of international security of states in the Post World War II international order was intended to cover not only political and military security but also the economic security of states and of their population.

The LG&E tribunal argued that “when a state's economic foundation [is] under siege, the severity of the problem could equal that of any military invasion”.⁹²

In the Deutsche Telekom and CC/Devas disputes, which were raised after the Argentine cases, India, as the respondent in both cases, argued that its determinations on matters of national security were owed substantial deference and that international tribunals should not pre-empt national security determinations made by national authorities, as the latter were uniquely positioned to determine what constituted a state's essential security interests.⁹³ The tribunals in both cases agreed that the government should be given a degree of discretion in defining its own security interests.⁹⁴ In the CC/Devas case, the tribunal asserted that national security is tied to the existential core of a state and that it could not rule on national security issues.⁹⁵ In the Deutsche Telekom award, the arbitral tribunal clarified the limits of the state's discretion to determine its security interests, but did not find this deference unlimited.⁹⁶ By referring to the case law of the European Court of Human Rights (ECtHR), the tribunal held that even though national security had a broad scope, this notion could not be extended well beyond its natural meaning.⁹⁷ The ECtHR, in its interpretation of national security and whether drug trafficking could endanger national security, explained that:⁹⁸

It is true that the notion of ‘national security’ is not capable of being comprehensively defined. It may, indeed, be a very wide one, with a large margin of appreciation left to the executive to determine what is in the interests of that security. However, that does not mean that its limits may be stretched beyond its natural meaning.

In WTO law, two recent panels have acknowledged that it is within the state's discretion to define what a security interest is, although such a discretion is still subject to a good faith review.⁹⁹ In Russia-Traffic in Transit, the panel defined essential security interests as “[i]nterests related to quintessential functions of the state, namely the protection of its territory and its population from external threats and the maintenance of law and public order internally”.¹⁰⁰ The panel affirmed that the determination of security interests depends on the perceptions of the state concerned, and that these interests can be expected to vary with changing circumstances.¹⁰¹ However, the panel added that:¹⁰²

This does not mean that a Member is free to elevate any concern to that of an ‘essential security interest’. Rather, the right of a Member to designate particular concerns as ‘essential security interests’ is limited by its obligation to interpret and apply Article XXI(b)(iii) of the GATT 1994 in good faith. The Panel recalled that the obligation of good faith is a general principle of law and a principle of general international law which underlies all treaties, as codified in Article 31(1) and Article 26 of the Vienna Convention.

In Saudi Arabia – Intellectual Property Rights, the panel made further clarifications to the application of this standard of review. By referring to the requirement of good faith, it considered that

[t]he requirement that an invoking Member articulates its essential security interests sufficiently to enable an assessment of whether the challenged measures are related to those interests is not a particularly onerous one, and is appropriately subject to limited review by a panel.¹⁰³

Thus, it can be seen that recent practice of WTO panels and investment tribunals have endorsed “a minimal satisfactory standard” and recognized that a degree of discretion is left to the state to determine “essential security interests”.

In the following discussion, we address the question of how to apply a good faith test in such a way as to avoid abuse. To determine what situations the term “security interests” encompasses, it is not logical to recognize a general, universal approach that should be applied to all treaties in all circumstances. What it requires, though, is that in each case the arbitral tribunal must consider all the circumstances and conditions on a treaty-by-treaty basis and adopt the appropriate standard of review by applying the rules of treaty interpretation. In other words, standards of review need to emerge endogenously out of the interpretation of concrete legal texts and not be imposed from the outside.¹⁰⁴ As such, this section is confined to analysis of US treaty practice with respect to the definition of “security interests”. Meanwhile, in light of this, some conclusions can be reached on GATT-type clauses.

The approach advocated in this article privileges a textual interpretive framework in order to determine “security interests”. This approach seeks to ascertain how words are ordinarily used in international economic law. Particularly relevant in our analysis are other international treaties concluded by the contracting parties in the field of investment.¹⁰⁵ On this basis, one could suggest that treaty terms are interpreted according to the meaning they possess “in the light of current linguistic usage, at the time when the treaty was originally concluded”.¹⁰⁶ But actual linguistic usage may not always be decisive and conclusive. As Dawidowicz rightly puts it, “account must sometimes be taken of those instances in which the meaning of a term has evolved over time”.¹⁰⁷ International tribunals have also affirmed that treaty terms are not static, but, by definition, are evolutionary.¹⁰⁸ According to such an approach, this part tries to investigate the ordinary meaning of security interests found in the security exception clause of the US Model BIT by scrutinizing the usage and historical origin of this objective.

Early US FCN treaties that included a security exception clause addressed national interests in war situations. For instance, the US–Austria FCN Treaty in 1931, provided that, “in the event either High Contracting Party shall be engaged in war, it may enforce such import or export restrictions as may be required by the national interest”.¹⁰⁹ Such clauses with identical terms started to appear as regular provisions of US FCN treaties in the aftermath of the Second World War. These treaties expanded the realm of security interests, but at the same time these interests were related to political and military emergencies. In this regard, one good example is the US-Iran FCN treaty stating that:¹¹⁰

The present Treaty shall not preclude the application of measures: (a) regulating the importation or exportation of gold or silver; (b) relating to fissionable materials, the radio-active by-products thereof, or the sources thereof; (c) regulating the production of or traffic in arms, ammunition and implements of war, or traffic in other materials carried on directly or indirectly for the purpose of supplying a military establishment; and (d) necessary to fulfill the obligations of a High Contracting Party for the maintenance or restoration of international peace and security, or necessary to protect its essential security interests.

As can be inferred, security interests must be interpreted in light of these subparagraphs. Clearly, these subparagraphs are related to political and military conflicts and do not include other aspects of security, especially economic or environmental crises. The security clause was also adapted as part of the original GATT agreement signed in 1947. During the drafting process of Article XXI, the preparatory delegates, in which the US played a major role, acknowledged that the exception gave latitude for “security as opposed to commercial purposes”.¹¹¹ Moreover, some scholars who analyzed this article opined that it:¹¹²

address[es] the immediate political-military conditions that a state deems important for its position in the world … [and thus it] would be wrong to read article XXI as coping with dire socioeconomic consequences ensuing from the operation of the GATT principles and policies.

This understanding of the security exception provision is grounded in the widely shared concern over the abuse of Article XXI of the GATT.

In a similar vein, it is worth noting that the US explained the meaning of security interests in some BIT letters of submittal to the Senate. In this respect, in its letter of submittal for the Lithuania BIT to the Senate, it provided that “[m]easures permitted by the provision the protection of a Party's essential security interests would include security-related actions taken in time of war of national emergency”.¹¹³ An identical explanation on “security interests” also appears in the US-Ukraine BIT.¹¹⁴ As Moon highlights, while this definition does not control the interpretation of other treaties, it does illuminate certain linguistic expectations attached to “essential security”.¹¹⁵

While the ICJ in the Nicaragua case stated that “the concept of essential security interests certainly extends beyond the concept of an armed attack”,¹¹⁶ that is far from construing the language of essential security to include economic necessities.¹¹⁷ Notwithstanding this, the Court in Oil platforms did refer to the uninterrupted flow of maritime commerce as being a reasonable security interest of the US. However, it should be aware that such commercial interests were only considered relevant because armed attacks were at play. As Moon put it:

the broad interpretation of ‘security interests’ in the Argentine cases is likely driven by a consequentialist approach to treaty interpretation. It blurs the boundaries between legal doctrines and runs counter to the canonical rule of the VCLT to give effect to each provision in a treaty.¹¹⁸ A consequentialist approach to treaty interpretation also produces serious negative externalities, in the form of discouraging foreign investors’ interpretations.¹¹⁹

Based on this background, it could be said that, from the US's view, the ordinary meaning of “security interests” includes political and military emergencies. This conclusion is confirmed by considering the historical origin of the “security exception clause” in FCN treaties, and the manner in which the US has used these interests in other international instruments. But it should be kept in mind that, in the practice of international tribunals, much more emphasis is placed on the nexus requirement; that is, the necessity of the measure, with a very wide measure of discretion given to states to determine whether or not a matter relates to national security. That is because these interests are related to the sovereignty and existential core of the state, and it is exceedingly rare for an ad hoc tribunal to find that a particular interest considered by a state to be a security interest may not actually be recognized as such. Such a process of reasoning is found in recent jurisprudence of WTO panels and investment tribunals. For instance, the Panel in Saudi Arabia – Intellectual Property Rights explained the reason for applying such a limited review as follows:¹²⁰

This analytical step serves primarily to provide a benchmark … and enables an assessment by the Panel of whether either of the challenged measures found to be inconsistent with the TRIPS Agreement is plausibly connected to the protection of those essential security interests.

Such unjustified reasoning should be avoided or at least complemented by giving importance to the ordinary meaning of “security interests” as argued above. Under a good faith test applicable to the permissible objective, states are to be given a margin of discretion as to what constitutes a threat to their own national security, though this discretion ought not to allow states to invoke essential security interests in times of socio-economic emergencies. In fact, states do not have an unlimited power to elevate any concern to that of an “essential security interest” and to utilize the security exceptions as permanent loopholes to evade treaty obligations. In doing so, tribunals could take the historical origin and state practice of interpreting “security interests” into account based on a textual approach to determine whether a state acted in good faith to raise a particular interest as a security matter.

C. Nexus Requirement: Necessity

Another controversial element of security exception clauses is the nexus requirement, or the necessity of state measures to protect essential security interests. Under this element, there must be a link between the measures adopted by a state and its security interests. As argued above, the adjectival clause “it considers necessary” qualifies this element and it is up to the state's discretion to determine which measures are necessary; consequently, arbitral tribunals are not empowered to substantially review the measures. However, this does not mean that the state is exempted from adjudicatory oversight of regulatory measures. In jurisprudence of different legal regimes of international law as well as doctrine, it is generally accepted that the standard of review for examining this element is the good faith standard based on Article 26 of the VCLT. That is to say that the tribunal will analyze whether the state acted in good faith when adopting measures to tackle its perceived security threats. The principle of good faith as one of the most important and fundamental principles of international law has no clear and unambiguous definition, which is why Bin Cheng has acknowledged that “what exactly this principle implies is perhaps difficult to define … [it] eludes a priori definition”.¹²¹ Other scholars have stated that this principle can be illustrated but not defined.¹²²As international tribunals have not adopted a workable approach for applying it, they will have to develop their own approaches to decide whether the good faith requirement has been met.¹²³ So, we, like Kolb, believe that good faith as a legal concept is not a univocal or univalent concept. It is a multiple concept with different functions depending on the context.¹²⁴

In consideration of this introduction, it is clear that the principle of good faith is a general concept and, therefore, it is necessary to identify a practical standard for applying the principle to specific situations. In this matter, the question that arises here is which approach can operationalize the good faith review in international investment law. Some tribunals and commentators have suggested divergent approaches in the application of good faith as a standard of review.

1. International Jurisprudence

In the Mutual Assistance case, the ICJ has addressed the good faith obligation in the application of a self-judging clause. In order to establish a good faith test, the Court ruled that France must show that the reasons for refusing the request and execution of the letter rogatory fell within those mentioned in Article 2 of the Convention.¹²⁵ The Court found that one of the reasons for the French judge to refuse the request was the presence of declassified secret defence documents related to the security interests of the country, which are covered by Article 2. The ICJ concluded that France had relied on Article 2 in good faith.¹²⁶

As scholars have observed, the ICJ “interpreted good faith to permit only a very limited review”¹²⁷ that “resembles a touch and feel type test”¹²⁸ as the Court did not consider the factual situation of the case for its analysis but merely relied on the letter rogatory to find a security reason for the refusal. Therefore, the criteria for concretizing good faith standard remained unanswered in the court's judgment. However, Judge Keith, in his separate declaration, provided further clarifications for applying good faith. He noted that the principles of abuse of right and détournement de pouvoir were relevant in the exercise of a state's discretion. In his opinion, those principles required the state agency in question to exercise power for the purposes for which it was conferred, without regard to improper purposes or irrelevant factors.¹²⁹ He argued that in the words of the ICJ in the Gabcikovo-Nagymaros case, the good faith obligation reflected in Article 26 of the VCLT “obliges the Parties [to a treaty] to apply it in a reasonable way and in such a manner that its purpose can be realized”.¹³⁰

Two recent WTO panels tasked with the interpretation of security exception clauses have provided an appropriate approach for investment tribunals. In Russia-Traffic in Transit, the panel's report considered that the good faith obligation has to be concretized by employing a minimum requirement of plausibility. The panel stated that “this obligation is crystallized in demanding that the measures at issue meet a minimum requirement of plausibility in relation to the proffered essential security interests, i.e. that they are not implausible as measures protective of these interests”.¹³¹ The panel had to engage in determining whether the measures were so remote from, or unrelated to, the 2014 emergency.¹³² Therefore, in the panel's view, if there is no meaningful and pertinent relationship between the measures and the end, the state does not meet the good faith requirement and the element of necessity.

By applying this standard of review, the panel explained that, in 2014, when there was an emergency in Russia's relations with Ukraine that affected the security of the Ukraine-Russia border, the UN General Assembly recognized that this involved armed conflict.¹³³ It concluded that the transit restrictions taken by Russia in 2014, were correlated responses required to address an emergency situation.¹³⁴ In Saudi Arabia-Intellectual Property Rights, the panel that determined whether the Saudi Arabia measures were in connection with the security interests of the country employed more precise legal terms to elucidate the good faith test and “the minimum requirement of plausibility”. It stated that this standard means there must be a logical and rational link between the measures and security interests.¹³⁵ In this regard, the panel found that, given that there was a situation of heightened tension or crises between the parties, and that Saudi Arabia imposed a travel ban on and an expulsion order for all Qatari nationals to protect its security interests (to protect the state from the threats of terrorism and extremism), there was a plausible link between these interests and the measures preventing “beIN” from obtaining Saudi legal counsel to enforce its intellectual property rights through civil enforcement procedures before Saudi courts and tribunals.¹³⁶ In the panel's view, however, this connection did not exist between Saudi Arabia's stated interests and its authorities' non-application of criminal procedures and penalties to “beoutQ”.¹³⁷

2. Scholarly Approaches

Some commentators have suggested legal approaches to elucidate the content of good faith analysis in relation to self-judging clauses. The first approach suggested is the proportionality test. The proportionality test in international law applies to countermeasures and self-defence, and means that a state's reaction should never be greater than the original attack or wrong.¹³⁸ Some scholars have advocated its application to the necessity element by illustrating that in obvious cases where security is not at stake, or in which a measure will have no effect on protecting the interest, or is clearly not suited to reach the acclaimed goal, or in which the harm caused by the measure is excessively greater in relation to the intended security aim, the corresponding state action cannot be justified under the security exception clause.¹³⁹

The proportionality test has been routinely applied to the general exceptions – which are of non-self-judging nature – for determination of the necessity of measures in the context of EU law. Despite the discretion granted to the member states in Article 346 of the TFEU, the European Court of Justice (ECJ) in Commission v. Spain referred to this test when analyzing self-judging security exception and held that:¹⁴⁰

It is for the Member State which seeks to rely on those exceptions to furnish evidence that the exemptions in question do not go beyond the limits of such cases. In the present case, the Kingdom of Spain has not demonstrated that the exemptions provided for by the Spanish Law are necessary for the protection of the essential interests of its security. It follows that the VAT exemptions are not necessary in order to achieve the objective of protecting the essential interests of the security of the Kingdom of Spain.

Therefore, this judgment showed that the ECJ did not limit itself to mere control of outright abuses but, rather, made its decision by taking into consideration the proportionality on whether the measure in question was actually necessary for the protection of the state's security interests.

The application of this test, in our opinion, goes beyond a good faith review and does not consider the explicit formulation of the security exception clause. It should be noted that the cases in which there is no risk for security interests, or the measures are taken for commercial objectives, are evidently out of the scope of the self-judging provision, but their legal basis, as we will address later, cannot be drawn from proportionality. On the one hand, nothing in the text of the treaty requires the proportionality of a state's measure. On the contrary, it is explicitly left to the state to determine the necessity of measures. Put simply, a tribunal is not authorized to replace the state's own determination by finding that the adopted measures were not proportionate to achieve the objective pursued.¹⁴¹ On the other hand, as Kurtz noted, “the proportionality test often involves the weighing of complex value-laden and empirical judgements, it is highly doubtful that courts (tribunals) will be able to assess them adequately”.¹⁴² Likewise, the proportionality test has been applied to the non-self-judging clauses in EU and WTO law, and it is obvious that in the case of the self-judging security exception clause there is no basis for its application, and it would be considered inconsistent with the rules of treaty interpretation and the intention of the parties. Furthermore, if the ECJ referred to this test to analyze Article 346, that will be due to specific EU legal rules and the permission of Article 348 of the TFEU, which expressly allows the ECJ to review measures taken by the member states where improper use is being made of the power to protect security interests.¹⁴³

Akande and Williams suggested a subjective test as an alternative approach for the application of the good faith review, according to which the wording used in such a clause signifies whether the measure taken is necessary: it is not left to objective determination but, rather, to the subjective determination of the state.¹⁴⁴ In their view, a good faith test means that the member invoking the self-judging security exception clause must genuinely – or in fact, or subjectively – consider that there is some threat to its security interests.¹⁴⁵ In other words, the tribunal must determine whether the state in fact considers that the measure is necessary¹⁴⁶ – but it is not empowered to go beyond this level of review. To justify their opinion, in addition to the terms of the self-judging clause, these scholars mention the notion of abuse of rights pursuant to which a state may not misuse its right to take certain actions for purposes which go beyond the purpose for which the right exists.¹⁴⁷ But this test of bona fide belief might be impractical because it is very difficult for the other party, an investor, to prove that the state did not really consider the measure to be a necessary one. Although commentators have argued that this question relates to proof and evidence, and that there might well be evidence to suggest that the measure was not taken for security reasons, the question is how a private investor can access documents concerning state decisions on security interests that are highly sensitive. Besides, as scholars of international law have stated, good faith has a wider realm than a subjective meaning, thus we should not limit the concept of good faith to this narrow aspect.¹⁴⁸ As we will explain below, the objective aspect of good faith might be more important when considering an applicable test to uphold the necessity element. Also, as we are exploring a legal test to concretize the good faith standard, taking such a subjective approach clearly cannot help achieve this purpose.

Another approach for applying the principle of good faith is advanced based on comparative administrative law and discretionary powers of administrations.¹⁴⁹ In a detailed study, Schill and Briese analogize the state's discretion in the case of a self-judging clause to the discretionary powers of administrative agencies in domestic law and state that there is no fundamental difference in the judicial review of both cases.¹⁵⁰ As they explained, in the domestic law of most countries a genuine domain is granted to administrative agencies in which they can exercise their discretion when administrating complex programmes.¹⁵¹ However, this discretion is subject to judicial review, and restrictions have been imposed by legal systems to prevent abuse of power.¹⁵² Having clarified approaches of the civil and common law, Schill and Briese conclude that despite differences in the scope and extent of judicial review in these countries, they all encompass formal and procedural control mechanisms rather than the review of the substance of the decision itself.¹⁵³ The authors consider these formal controls to be applicable to the self-judging element of the security exception clause in order to prevent state abuse and, in so doing, the court must review whether the factual basis of a state's decision was adequate and properly investigated, whether the appreciation of the governing legal framework was correct, whether the state abided by the proper procedure, and whether it was guided in the exercise of discretion by relevant and pertinent considerations in view of the purpose of the treaty in question.¹⁵⁴ A negative response to one of these questions means that the state has not met the standard of good faith review.

This view neglected to consider a judicial review of the connection between the contested measure and the claimed goal because, as commentators have stated in their studies, in most countries – in addition to the above formal and procedural restrictions – the judicial organ examines the decision of the agency in a substantial manner. In some countries, such as the UK, Australia, Canada, and the US, this review includes the existence of a reasonable link between the decision and the public end,¹⁵⁵ while in others, such as France, it is a proportionality test.¹⁵⁶ Further, in a case where the state did not respect one of these formal restrictions in coming to its decision and the state's measure was, in fact, necessary and the only way to protect its security interests, one may question whether, in such a situation, it can be concluded that the state did not act in good faith. The logical answer to this question would be in the negative.

In our view, the proper standard of review applying to the nexus element of a self-judging security exception clause should be sought in the concept of good faith. As international law scholars have pointed out, good faith has three aspects.¹⁵⁷ In the subjective sense, good faith is thought to be nothing more than:¹⁵⁸

a psychological concept, making reference to some sort of inner knowledge and correctness … it seems to be a properly moral concept, searching to introduce into the law a measure of rectitude, correctness, fair dealing, honest belief, absence of malice.

In order to apply this aspect, we must analyze whether the state – in fact and subjectively – deemed the measures to be taken were necessary. Accordingly, if the state considered its actions necessary, but these actions had been ineffective or less effective, the state still holds an honest belief and has met the good faith requirement. Another side of good faith is its objective aspect. In this regard, good faith is a general principle that protects certain finalities anchored in the common interest against excessive individualist pretenses (the prohibition of abuse of rights).¹⁵⁹ It also seeks to “concretize” its legal contents by reversing them into a series of intermediary principles like prohibition of abuse of rights, détournement de pouvoir, and fraud to the law.¹⁶⁰ The approach taken by Judge Keith in the Mutual Assistance case is consistent with this aspect of good faith. Therefore, from this perspective, the state's actions must be adopted to protect the permissible objective stated in the security exception clause of the treaty, and if these were adopted for another purpose, such as economic and commercial goals, invoking the clause is just a pretext to escape from treaty obligations.

Finally, good faith is sometimes regarded as a standard of review, by reference to which we assess the correctness of the behavior of the state in that matter.¹⁶¹ Here, good faith means reasonableness¹⁶² and is considered to function as a “logical and rational link” between the actions of a state and the legal norm upon which those actions are based.¹⁶³ In this light, there must be a logical link between the adopted measures (means) and pursued ends so that any reasonable person (state) in the state's position would have adopted those measures to protect its national security. It is believed that in the case where the tribunal intends to concretize good faith and find the proper standard of review to apply, this is the third aspect that must be adopted – because that is what good faith means as a standard. By rational or reasonable link, it is not meant that a substantial or causal relationship must be established, but what is required is that there must simply be a tenuous link between the adopted measures (means) and pursued ends so that the state's action must be “demonstrably rational”; that is, any reasonable person (state) in the state's position would have adopted those measures to protect its public interests. Such a criteria is not without precedence in international jurisprudence. As examined above, WTO panels have adopted “a minimum plausible means-end relationship” as regards the nexus requirement of self-judging security exception clauses. This approach would be most relevant, not least because states have somewhat bluntly grafted WTO-style exceptions onto their IIAs through the reproduction of the texts of Articles XXI of the GATT. Therefore, it must be considered as a rational link, implying that the measures taken by the state must not be remote from or unrelated to the security interests.

IV. Conclusion

The central objective of the international investment law system is to protect foreign investors by providing guarantees and a stable legal framework via investment treaties. This suggests that great care must be taken to not deviate from this goal. In particular, allowing a state to unilaterally suspend treaty obligations by simply invoking the self-judging security exception clause will undermine this legal system.

Although the self-judging clause gives states a wide scope of discretion compared to the non-self-judging clause, it does not oust the judicial review of international tribunals. This article has sought to provide the proper interpretation of the self-judging security exception clause consistent with the rules of treaty interpretation, and determined which approach may avoid abuse of such clauses. It has been demonstrated that unequivocal wording such as “it considers necessary” is required to identify the self-judging character. This is because, on the one hand, such a clause is exceptional and extraordinary and, on the other, the subject, purpose, and context of investment treaties is to support foreign investors. It is acknowledged that a self-judging clause does not affect the jurisdiction of the tribunal, but it limits judicial review in the merits phase to good faith review, as codified in Article 26 of the VCLT.

As for developing an appropriate standard on interpreting the provision, the analysis has shown that it generally consists of two main elements. Regarding the first element, namely security interests, we suggest that the determination of these interests should be given to the states themselves. Nevertheless, they must act in good faith and limit their determination to the ordinary meaning of the security interest in light of its historical origin and state usage. In the interpretation of the necessity element that is self-judging, even though there is no doubt that the applicable standard of review is good faith, scholars have divergent approaches on how this principle can be concretized. In our view, those approaches have been made without regard to the concept and aspects of good faith. It has been suggested that this standard can be operationalized by applying its three aspects: honest belief; the prohibition of abuse of rights; and, more importantly, the existence of a reasonable link between the measure and the aimed goal which is, here, the preferred and proper view.