Europe and the World: A law review Autonomy and international investment agreements after Opinion 1/17

The application and implications of the principle of autonomy for international investment agreements concluded by the Member States and the European Union (EU) has become a recurrent theme before the Court of Justice of the European Union. The decisions in Achmea and Opinion 1/17 show that autonomy unfolds differently in intra- and extra-EU investment relations and can only be preserved in the latter context. The present article examines this difference and, in light of Opinion 1/17 , seeks to explain how and why the autonomy of EU law can be preserved for international investment agreements through careful treaty design. In addition, it sheds some light on the practical consequences for the EU’s and the Member States’ external investment relations.


Introduction
The outcome of Opinion 1/17 is inherently pragmatic. It enables the Commission to continue its role as a driving force of the MIC project, while simultaneously tying the EU's and the Member States' participation in investment dispute settlement mechanisms to a set of minimum standards safeguarding the autonomy of EU law. In doing so, Opinion 1/17 not only demonstrates that international (investment) law and EU law can harmoniously coexist, with the former developing and strictly observing the latter (Articles 3(5) and 21 Treaty on European Union (TEU)), 8 it further demonstrates how international investment law can be separated from EU law through careful treaty design, so that any interaction between the two (then independent) legal orders preserves the autonomy of EU law. As a result, Opinion 1/17 represents a change in the CJEU's autonomy jurisprudence that could enable the EU to become a more prominent player in -and an architect of -global investment law.
The present article examines the application of the principle of autonomy in the field of investment law and its implications for the design of international investment agreements (IIAs) by the EU and the Member States. Section 2 briefly recalls the conception of autonomy in light of the CJEU's past case law. Section 3 sheds some light on the first authoritative interpretation and application of the autonomy principle in the field of international investment law, that is, the Achmea case, and aims to explain why autonomy unfolds differently in intra-and extra-EU investment relations. Section 4 then turns to the EU's IIAs and Opinion 1/17. It analyses the procedural and material safeguards that were enshrined in the design of CETA and seeks to outline why these safeguards were deemed sufficient by the CJEU for preserving the autonomy of EU law. Section 5 highlights resulting practical implications for the design of IIAs by the EU and the Member States. Section 6 concludes.

The autonomy of EU law: A brief review of key developments and facets
The principle of autonomy 9 of the EU legal order is a judge-made legal construct. There is no reference to it in the Treaties. Autonomy first emerged in the early 1960s in the context of the Van Gend & Loos and Costa/Enel judgments, and is therefore closely linked to the construction of (then) Community law as a 'new legal order' 10 that 'may not be overridden by domestic legal provisions however framed, without being deprived of its character as Community law '. 11 At that point in time, autonomy had an internal dimension. It enabled (then) Community law to withstand challenges from national law by bolstering legal principles that establish a hierarchy of norms and foster the effective judicial protection of individual rights. These principles included, 'in particular', 12 the primacy and direct effect of (then) Community law. 13 It has been widely argued that autonomy has since met its internal objectives, with EU and national courts cooperating successfully to give full effect to the constitutional parameters of the 'new legal order'. 14 Yet, the already infamous decision of the German Federal Constitutional Court in PSPP, 15 which set aside the CJEU's decision in Weiss 16 and, contrary to the CJEU, concluded that the Public Sector Purchase Programme (PSPP) was ultra vires, shows that the primacy of EU law continues 8   to be contested in certain situations. It remains to be seen whether the internal dimension of autonomy will experience a revival. 17 That autonomy also unfolds vis-à-vis international law first became evident in Opinion 1/91, where the CJEU found that the agreement establishing the European Economic Area (EEA) had an 'unacceptable adverse effect on the autonomy of the Community legal system'. 18 It has since become the principle's main purpose to protect the systemic functioning, integrity and unity of the EU legal order against threats that arise from the EU's and the Member States' foreign relations. 19 The present article does not aim to examine the developments of (external) autonomy in great detail, nor does it aim to provide an overview of all of the principle's facets. 20 There are, however, some developments and facets of (external) autonomy that provide a useful backdrop for examining its role in the field of international investment law in general, and in the aftermath of Opinion 1/17 in particular.
To start with, it may be worth recalling that autonomy remains a 'nebulous' concept. 21 The CJEU has repeatedly emphasised that the EU is, in principle, competent to make agreements containing dispute settlement mechanisms. 22 Yet, in the same vein, the CJEU 'declared that an international agreement may affect its own powers only if the indispensable conditions for safeguarding the essential character of those powers are satisfied and, consequently, there is no adverse effect on the autonomy of the EU legal order'. 23 That statement is broad, and the number of 'indispensable conditions' open ended. Past case law showed that they include, inter alia, the allocation of competence between the EU and the Member States; the principles of unity, solidarity, loyal cooperation and mutual trust; the role and function of Member State courts as EU courts; the CJEU's exclusive jurisdiction to interpret and apply EU law; and the judicial protection of fundamental rights. 24 Autonomy therefore principally 25 safeguards the structural dimension of EU law. 26 Whether a specific agreement leaves the structural dimension and essential character of EU law intact can only be determined by the CJEU on a case-by-case basis, in accordance with the procedure stipulated in Article 218(11) TFEU. 27 It is unsurprising, therefore, that questions about autonomy sparked a long and controversial line of case law. Except for Opinion 1/92 (on the revised EEA agreement), Opinion 1/00 (on the European Common Aviation Area) and now Opinion 1/17, the CJEU found fault with the preservation of autonomy. In Mox Plant it held that the settlement of a dispute between the United Kingdom and 17 See F de Abreu Duarte and M Mota Delgado, 'It's the Autonomy (Again, Again and Again), Stupid!: Autonomy Between Constitutional Orders and the Definition of a Judicial Last Word', VerfBlog (2020) <https://verfassungsblog.de/its-the-autonomyagain-again-and-again-stupid/> accessed 19 June 2020. 18  The CJEU has repeatedly held that 'an international agreement providing for the creation of a court responsible for the interpretation of its provisions and whose decisions are binding on the European Union, is, in principle, compatible with EU law'. See most recently Opinion 1/17, para 106. 23 See e.g. Opinion 2/13, para 183. 24 Opinion 2/13, paras 159 et seq. 25 As will be seen further in Section 4.2, autonomy exceptionally also has a substantive dimension, and that substantive dimension was further developed by the CJEU in Opinion 1/17. Ireland by an arbitral tribunal set up under the United Nations Convention on the Law of Sea would impinge on its exclusive power to interpret and apply EU law. 28 In Opinion 1/09 the CJEU concluded that a European Patents Court would deprive Member State courts of their task to refer questions for preliminary rulings. 29 Most controversially, 30 the CJEU declared the draft agreement providing for the EU's accession to the European Convention on Human Rights (ECHR) incompatible with EU primary law in Opinion 2/13. 31 Drawing heavily on that Opinion, the CJEU found in Achmea that intra-EU BITs remove disputes from Member State courts and impinge on its exclusive jurisdiction to interpret and apply EU law. 32 In this past case law, the justifications for a breach of the autonomy principle often linked back to the CJEU's own exclusive jurisdiction, either directly (especially Article 344 TFEU) or indirectly (especially the role of national courts as EU courts (Article 19 TEU), and their ability to make preliminary references (Article 267 TFEU)). The ostensible eagerness of the CJEU to shield its own jurisdiction from other international courts and tribunals led Bruno de Witte to conclude that the CJEU 'has occasionally been a little selfish, showing more concern for its own role than for the advancement of the broader agenda of promoting international law'. 33 Jan Klabbers found it 'worrisome' that the CJEU 'aspired[d] to build a fence around EU law, therewith running the risk of placing the EU outside international law'. 34 Piet Eeckhout argued that the CJEU used 'a concept of the autonomy of EU law which borders on autarky'. 35 Ramses Wessel and Steven Blockmans nevertheless noted that EU law 'perhaps even paradoxical' displays of a certain openness towards international law, which 'implies the acceptance of [international] norms on the EU legal order'. 36 And the CJEU's past case law presented some indication that autonomy must not be construed as 'a synonym for autarchy', 37 as Advocate General Bot put it, but that it may be reconciled with the EU's striving for international openness, including in regard to (other) international courts and tribunals. In reference to Opinions 1/91 and 1/92, the CJEU found in Opinion 1/00 that international courts or tribunals 'must not have the effect of binding the EU and its institutions, in the exercise of their internal powers, to a particular interpretation of the rules of EU law' 38 -a statement that it has since frequently recalled. 39 This statement suggests that autonomy does not per se prevent an international court from assessing EU law. 40 Instead, it appears to imply that international courts, when interpreting the relevant international treaty norms, must neither issue a legally binding interpretation of EU law, nor deliver a decision that affects the operation of the EU institutions in accordance with the EU's constitutional framework. As will be shown further below, that is precisely the definition of autonomy   40 See also Koutrakos, n 7, 4-5.
Opinion 1/17 builds upon (Section 4). In linking autonomy back to its original construction, Opinion 1/17 might be viewed as an (attempted) rebuttal of the criticisms that arose especially in the aftermath of Opinion 2/13 and, most recently, Achmea.

A context-specific threshold of protection? Autonomy, Achmea, and intra-EU BITs
In Achmea the CJEU was for the first time directly asked to rule on the compatibility of an IIA with EU law. The case concerned a Member State inter se agreement, namely a BIT between Slovakia and the Netherlands (a so-called intra-EU BIT). The CJEU concluded that the ISDS mechanism of the BIT breached both Articles 267 and 344 TFEU. 41 The Achmea decision was criticised for portraying a 'maximalist position' on autonomy. 42 This is especially reflected in the criteria used by the CJEU to assess interactions between international investment law and EU law. It bears noting that it is the existential task of international courts and tribunals to judge domestic legal measures against international treaty standards. Article 8 of the Dutch-Slovakia BIT accordingly stipulated that the ISDS tribunal should take account of, inter alia, 'the law in force of the Contracting Party concerned' and 'the provisions of this Agreement'. Unlike CETA, the BIT did not expressly exclude EU law from the applicable law interpreted by the arbitral tribunal (see further Section 4.1). However, it is arguably difficult to establish on the basis of Article 8 of the BIT alone whether the ISDS tribunal may interfere with the CJEU's monopoly to give a definite interpretation of EU law. As explained in Section 2, the definition of autonomy that emerged from Opinions 1/91, 1/92 and 1/00 does not focus on the interaction of international and EU law as such, but on the effects of such an interaction on the constitutionally allocated powers of the EU institutions. In Achmea, the CJEU, however, appeared to find the interaction of international and EU investment law sufficient for the autonomy of EU law to be at stake. It considered the intra-EU ISDS tribunal to be liable to relate to EU law. 43 The CJEU therefore solely assessed the potential and not the actual scope of disputes brought before the intra-EU ISDS tribunal. 44 It questioned whether EU law pertains to the dispute at issue, and not whether the intra-EU ISDS tribunal's assessment of EU law might bind the EU institutions, including the CJEU, to a specific interpretation of EU law. 45 This protective threshold would, in practice, be difficult to meet by any international court or tribunal, unless it were firmly embedded in the EU's judiciary system and able to make preliminary references to the CJEU. It is, however, the main purpose of many international courts and tribunals, including in the field of international investment law, to establish an alternative dispute settlement route that is not embedded in the constitutional system of one of the parties to the dispute. 46 The Achmea case hence suggests that Articles 267 and 344 TFEU may be triggered as soon as an international court or tribunal is not part of the EU judicial system and interacts -in whatever form -with EU law. In this regard, the conception of autonomy that emerged from Achmea links back to Opinion 2/13, where the CJEU held that that there is a 'necessity for the prior involvement of the Court of Justice in a case . . . in which EU law is at issue . . . '. 47 41 Art 344 TFEU expressly only relates to Member States and it was therefore questioned, including by the Bundesgerichtshof, whether the provision could be applied to disputes between Member States and individuals. Achmea suggests that a joint reading of arts 267 and 344 TFEU implies that the CJEU's exclusive jurisdiction expands to disputes between Member States and individuals (para 60). In Opinion 1/17 the CJEU somewhat detached the principle of its own exclusive jurisdiction from art 344 TFEU, stating that international courts must generally not have power to interpret and apply EU law (para 199). Art 344 TFEU is not mentioned in Opinion 1/17. 42 Koutrakos, n 7, 8. 43 Case C-284/16, Achmea, para 39. 44 See also Eckes, n 7, 85. 45 See also Gatti, n 7, 116. 46 The idea behind ISDS/ICS is to create a system for investment adjudication that functions independently from domestic courts. According to the UNCTAD World Investment report, '[t]he ISDS mechanism was designed to depoliticize investment disputes and create a forum that would offer investors a fair hearing before an independent, neutral and qualified tribunal. . . . Given that investor complaints relate to the conduct of sovereign States, taking these disputes out of the domestic sphere of the State concerned provides aggrieved investors with an important guarantee that their claims will be adjudicated in an independent and impartial manner', 111-12. <https://unctad.org/en/PublicationsLibrary/wir2013_en.pdf> last accessed 30 September 2019. 47 Opinion 2/13, para 237, emphasis added.
Several commentators argued that if the CJEU were to extend its reasoning in Achmea to other international agreements, CETA's ICS should not pass the protective threshold of autonomy. Irrespective of any 'autonomy safeguards' enshrined in CETA, the treaty would not fully prevent issues pertaining to EU law from being assessed by an international tribunal that cannot make preliminary references to the CJEU. 48 Opinion 1/17, however, shows that the threshold for triggering the CJEU's exclusive jurisdiction that emerged from Achmea is confined to the specific context of intra-EU BITs. And there are arguably good reasons for such a narrow and context-specific reading of Achmea. 49 In Achmea the CJEU emphasised that the autonomy of EU law can only be preserved if the EU's court system remains able 'to ensure consistency and uniformity in the interpretation of EU law'. 50 This criterion might be especially vulnerable in the context of Member State inter se agreements. Intra-EU BITs postulate legal rights only for investors of certain Member States and establish an alternative dispute settlement system within the EU. Even if a normative conflict is unlikely (as Advocate General Wathelet argued in his Opinion in Achmea 51 ), it would, if it were to occur, inevitably pose a threat to the uniform interpretation of EU law because 'that law must be regarded both as forming part of the law in force in every Member State and as deriving from an international agreement between the Member States'. 52 The CJEU's high protective threshold for intra-EU ISDS tribunals assessing EU law may thus be explained by the specific constitutionalised system EU law establishes to govern Member State inter se relations and the specific threats intra-EU investment arbitration poses to that system. A context-specific reading of Achmea therefore implies that the judgment is not transferrable to EU external relations and, specifically, EU IIAs. Indeed, the CJEU already made a distinction between intra-EU BITs and EU IIAs in the final paragraph of the Achmea case. 53 In that paragraph it re-emphasised the specific role of mutual trust between Member States, which is called into question by intra-EU investment arbitration. The CJEU's conception of autonomy in Achmea may, therefore, as Advocate General Bot highlighted, 'have been primarily guided by the idea that the judicial system of the European Union, in so far as it is based on mutual trust and sincere cooperation . . . , is inherently incompatible with the possibility of Member States establishing, in their bilateral relations, a parallel dispute settlement mechanism'. 54 The CJEU confirmed in Opinion 1/17 that mutual trust is absent in EU external relations. 55 That absence of trust, as well as the 'reciprocal nature of international agreements and the need to maintain the powers of the Union in international relations' would generally justify the EU's participation in international dispute settlement systems. 56 Achmea and Opinion 1/17 therefore drew a clear dividing line between Member State inter se and EU agreements establishing dispute settlement mechanisms based on their different effects on the internal trust in, and uniform interpretation of, EU law on the one hand, and the external need to establish and enforce reciprocal legal standards on the other hand. This distinction may explain why EU IIAs, in contrast to intra-EU BITs, can preserve the autonomy of EU law through careful treaty design. 48 Eckes, n 7; and Gatti, n 7. 49 See also Koutrakos, n 7, 6 et seq; and Contartese, n 26. 50 Case C-284/16, Achmea, para 35. 51 The Advocate General relied on UNCTAD statistics, which showed that investors were only successful in 10 out of 63 intra-EU arbitral proceedings, only one of which (the Micula case) concerned the (mal-)application of EU law by a tribunal. He therefore argued that 'the systemic risk which, according to the Commission, intra-EU BITs represent to the uniformity and effectiveness of EU law is greatly exaggerated'.

Safeguarding autonomy through careful treaty design: Opinion 1/17 and EU IIAs
In contrast to intra-EU BITs, recent IIAs such as CETA consider the potential consequences of interactions between international investment law and EU law for the autonomy of EU law. The ICS established in CETA attempts to separate international and EU law, especially by confining the scope of the CETA tribunals' jurisdiction to the rules of CETA and by underlining that awards do not impinge on the powers of the EU institutions to protect public interest objectives in accordance with the EU Treaties. To this end, several 'autonomy safeguards' were inserted in CETA. In Opinion 1/17 the CJEU was essentially asked to determine whether these safeguards ensure the compatibility of CETA with the autonomy of EU law.
The CJEU used two general criteria to assess whether CETA is compatible with the autonomy of EU law: CETA must not (a) 'confer on the envisaged tribunals any power to interpret or apply EU law other than the power to interpret and apply the provisions of that agreement . . . '; and (b) 'structure the powers of those tribunals in such a way that . . . they may issue awards which have the effect of preventing the EU institutions from operating in accordance with the EU constitutional framework'. 57 In contrast to Achmea, the CJEU therefore did not use mere interactions of international investment and EU law as a benchmark, but focused on the potential effects of these interactions on the interpretation and application of EU law and on the constitutional powers of the EU institutions. In this regard, the CJEU re-emphasised its original conception of autonomy that emerged from Opinions 1/91, 1/92 and 1/00 (see above Section 2).
In its decision in Opinion 1/17, the CJEU found that CETA fulfilled the first criterion through procedural 'autonomy safeguards' (Section 4.1) and the second criterion through material 'autonomy safeguards' (Section 4.2). In doing so, the CJEU established clear guidelines as to how the autonomy of the EU legal order can be preserved through careful investment treaty design. Opinion 1/17 will therefore have a strong practical impact on the making of IIAs by the EU and the Member States (see below Section 5).

Procedural autonomy safeguards
CETA aims to ensure that EU law cannot form part of the law interpreted and applied by the CETA tribunals. In a first step, CETA generally precludes investment courts from assessing the allocation of competence between the EU and the Member States in the process of determining the respondent party to an ICS dispute on the side of the EU. Pursuant to Article 8.21 CETA, it is for the EU to determine whether it will act as respondent, or whether it wishes to leave that role to a Member State ('rule of proceduralisation'). 58 The rule of proceduralisation addresses faults the CJEU found with the co-respondent mechanism in Opinion 2/13, which left discretionary powers with the European Court of Human Rights to rule on the vertical allocation of competence within the Union by reviewing the plausibility of either the EU or the Member States to serve as co-respondents. 59 Conversely, CETA does not foresee that the EU and the Member States act as co-respondents: it is for the EU to determine whether it wishes to act as a single respondent party or whether it wishes to leave that role to a Member State. 60 The CETA tribunals have no power to review the EU's decision. However, if the EU fails to give notice within 50 days, the respondent on the side of the EU will be determined in accordance with the rules set out in Article 8.21(4) CETA. 61 In its oral submissions, 62 Slovakia raised doubts about the compatibility 57 Opinion 1/17, para 119, emphasis added. of the rule of proceduralisation with EU law. It contended that after the 50-day period elapses, the EU respondent party -and thus the vertical allocation of competence within the Union -would be determined by international law. Yet, as was stressed by the Commission and Council, the rule of proceduralisation could not practically work without specifying a time limit within which the EU must inform the investor as to whether the EU or a Member State shall act as respondent. 63 Otherwise, the EU could (indefinitely) stall the investor's submission of a claim. Indeed, the CJEU seemed unimpressed by Slovakia's argument and did not address it in detail. It merely noted that the '[CETA] Tribunal cannot take notice of the division of powers between the European Union and its Member States. The difficulty identified by the Court in [inter alia] Opinion 2/13 . . . does therefore not arise in this case.' 64 Besides the 'rule of proceduralisation', CETA contains a second key 'procedural autonomy safeguard', namely its 'applicable law clause'. Pursuant to Article 8.31(1) CETA, the tribunal may only apply the provisions of CETA ('the applicable law'), read in light of the Vienna Convention on the Law of Treaties (VCLT) and other principles of international law applicable between the parties. The CETA tribunal shall have no jurisdiction to determine the domestic legality of a measure. 65 In determining whether a measure of the EU or Canada complies with the material standards set in Sections C and D of CETA, 66 the CETA tribunal may only 'consider, as appropriate, the domestic law of a Party as a matter of fact'. 67 In doing so, it shall follow the prevailing interpretation given to domestic law by the courts or authorities of a party and thus, for the EU, especially the CJEU. 68 Similarly, the CETA appellate tribunal cannot interpret or apply domestic law; yet, it may identify 'manifest errors in the appreciation of the facts, including the appreciation of relevant domestic law'. 69 CETA therefore takes greater care in separating the jurisdictions of the CJEU and the CETA tribunals than the draft agreement on creating a European Patents Court (Opinion 1/09) or the Dutch-Slovakia BIT (Achmea). The latter two agreements expressly designated EU law as 'applicable law' and thus attempted to directly integrate EU law into international law. CETA also differs from the draft agreement on the EU's accession to the ECHR (Opinion 2/13) in this respect as the ECHR and Charter of Fundamental Rights of the European Union ('the Charter') stipulate equivalent and interlinked standards of protection (Article 53 ECHR and Article 53 of the Charter). Given the clear separation of CETA from EU law, there was little disagreement in academic literature that the 'applicable law clause' significantly reduces the likelihood of CETA tribunals rendering a substantive interpretation of EU law. 70 But does the designation of law as a 'matter of fact' prevent the CETA tribunals from interpreting and applying EU law? This question was possibly one of the most contested aspects of Opinion 1/17. Courts have a margin of discretion when reading the law. It is therefore doubtful whether law can be factually 'considered' or 'appreciated' without simultaneously being 'interpreted'. After the CJEU delivered its judgment in Achmea, it was argued that CETA could at least not prevent disputes before the ICS potentially pertaining to EU law (see above Section 3). Proponents of this view further underlined that the CETA appellate tribunal may identify manifest errors in the supposedly factual assessment of domestic law. Therefore, the prior involvement of the CJEU was considered by some to be necessary for CETA to be compliant with EU law. 71 Opinion 1/17, however, confirmed that the prior involvement of the CJEU is not needed for CETA because international and EU investment standards form (sufficiently) independent bodies of law. CETA is not directly effective, implying that the agreement may not be directly invoked before EU courts (Article 30.6(1) CETA). By inference, EU norms may not be invoked before the ICS. Article 8.31 CETA emphasises this strict jurisdictional dividing line by stressing that the CETA tribunals may only interpret the rules of CETA in light of the VCLT and not consider the domestic validity of EU law -a task preserved exclusively for the CJEU. Indeed, the domestic legality of a measure should not be of any interest to the CETA tribunals. They assess whether a concrete measure by the EU or a Member State affects the rights of a Canadian investor enshrined in CETA, irrespective of whether that measure was prescribed by, in conformity with, or in breach of, EU law. 72 In this regard, ICS tribunals consider the EU or Member State measure as a matter of fact, as the ICS tribunals do not need to assess the validity of EU law to determine the conformity of a measure with the rules of CETA. As a consequence, the CJEU held in Opinion 1/17 that it is 'consistent that the CETA makes no provision for the prior involvement of the Court' because the 'CETA Tribunal and Appellate Tribunal stand outside the EU judicial system and . . . their powers of interpretation are confined to the provisions of the CETA'. 73 As regards the CETA appellate tribunal's power to overturn awards rendered by the tribunal based on manifest errors in the appreciation of domestic law, the CJEU found that it 'is nonetheless clear from the preceding provisions that it was in no way the intention of the Parties to confer on the Appellate Tribunal jurisdiction to interpret domestic law'. 74 The CJEU does not clarify why 'the intention of the parties' is a sufficient criterion for preserving the autonomy of the EU legal order. In view of the CJEU's above-mentioned criteria, it would arguably have been more consistent to focus on the power of the arbitral tribunal to bind the EU institutions, directly or indirectly, to an external interpretation or application of EU law. In that regard, it bears noting that -as both the Commission and Council emphasised in their oral submissions -awards are limited to monetary damages and/or the restitution of property. CETA excludes specific performance (Article 8.39(1) CETA). This generally permits applying an EU measure internally even if the CETA tribunals consider that measure in breach of CETA, at the cost of paying compensatory damages. Indeed, the CJEU noted that the CETA tribunals have no power to 'either annul the contested measure, or require that the domestic law of the Party concerned should be rendered compatible with the CETA'. 75 However, as the next section shows, it carefully considered whether ICS awards may affect the EU institutions' power to achieve public interest objectives and required that EU IIAs enshrine 'material safeguards' to that end.

Material 'autonomy safeguards'
Although CETA de jure excludes specific performance remedies, critics are concerned that the enforcement of ICS awards will de facto determine the course of public policy and public interest regulation by the parties. This concern was also raised by Belgium and Slovakia in Opinion 1/17. They argued that the CETA tribunal might be confronted with a situation where it has to balance the objective of an EU measure to preserve a public interest set out in the TEU and TFEU with the objectives of CETA. 76 As a result, the CETA tribunal might interfere with the power of the EU institutions to act in accordance with the Treaties and hence with the autonomy of EU law. The CJEU took that concern seriously. In particular, it held: The jurisdiction of those tribunals would adversely affect the autonomy of the EU legal order if it were structured in such a way that those tribunals might ... call into question the level of protection of a public interest .... If the Union were to enter into an international agreement capable of having the consequence that the Union -or a Member State in the course of implementing EU law -has to amend or withdraw legislation because of an assessment made by a tribunal standing outside the EU judicial system of the level of protection of a public interest established, in accordance with the EU constitutional framework, by the EU institutions, it would have to be concluded that such an agreement undermines the capacity of the Union to operate autonomously within its unique constitutional framework. 77 Although autonomy principally safeguards the structural dimension of EU law (see above Section 2), the integration of substantive elements into the protective cloak of autonomy is nothing new. The CJEU found in Kadi that the autonomous character of the (then) Community legal order implied that 'the obligations imposed by an international agreement cannot have the effect of prejudicing the constitutional principles of the EC Treaty, which include the principle that all Community acts must respect fundamental rights'. 78 Likewise, the CJEU was concerned that the ECHR may prescribe a different minimum standard of fundamental rights protection within the Member States than the Charter. 79 In this respect, the CJEU's conclusion that CETA may not affect EU public interest and fundamental rights protection is consistent with these previous findings.
The CJEU therefore had to assess whether CETA provided adequate 'material safeguards' to ensure that the ICS cannot call into question the level of protection of a public interest set by the parties. Article 8.9(1) CETA affirms the parties' 'right to regulate ... to achieve legitimate policy objectives, such as the protection of public health, safety, the environment or public morals, social or consumer protection or the promotion and protection of cultural diversity'. The parties may pursue their right to regulate 'in a manner which negatively affects an investment or interferes with an investor's expectations' (Article 8.9(2) CETA). The parties' right to regulate is reiterated in CETA's 'general exceptions' clause 80 and the Joint Interpretative Instrument 81 . Moreover, CETA contains a number of specific 'material safeguards' in the area of investment protection, such as an exhaustive list of cases where 'fair and equitable treatment' may be accorded 82 and a clause defining (indirect) expropriation and generally excluding measures preserving public interest objectives from constituting (indirect) expropriations. 83 For the CJEU, 'it is accordingly apparent from all those provisions contained in the CETA that ... the Parties have taken care to ensure that those tribunals have no jurisdiction to call into question the choices democratically made within a Party relating to, inter alia, the level of [public interest objectives] or, equally, fundamental rights'. 84 It therefore concluded that the material safeguards enshrined in CETA ensure that there are no effects on the operation of the EU institutions in accordance with the EU constitutional framework. CETA therefore does not adversely affect the autonomy of EU law. 85 By concluding that the CETA tribunals have no jurisdiction whatsoever to assess the parties' right to regulate, the CJEU appears to postulate that all measures directed at public interest objectives are carved out from the scope of CETA and the jurisdiction of the CETA tribunals. There is some irony to this conclusion: on the one hand, the CJEU emphasises that the CETA tribunals must not interpret and apply EU law in a legally binding manner. On the other hand, the CJEU seemingly sees no problem in assuming that it has the power to issue a legally binding interpretation of the scope of CETA and the jurisdiction of the CETA tribunals. Certainly, as the CJEU noted, the repetitive emphasis on the right to regulate in CETA gives the parties a broad freedom to determine the level of public interest protection, including through 77 ibid, paras 148, 150. 78 Joined Cases C-402/05 P and C-415/05 P, Kadi, para 285. 79 See esp. the remarks of the CJEU in regard to the 'insufficient' coordination between the standards of protection prescribed by arts 53 ECHR and 53 of the Charter set out in Opinion 2/13, paras 189-90. 80  the setting of new and more stringent standards that have a negative effect on investment liberalisation and protection. That does not, however, imply that all measures aimed at protecting public interest objectives are carved out from the jurisdiction of the CETA tribunals. The CETA tribunals may determine whether the impact of such measures is manifestly excessive or arbitrary. 86 The CETA tribunals may thus subject the parties' right to regulate to a limited proportionality analysis, preventing the parties from abusing their legitimate freedom to protect public interests as a disguised barrier to trade and investment. The CJEU's conclusion that the CETA tribunals have no jurisdiction whatsoever to call into question any type of measure aimed at protecting public interest objectives is thus imprecise, as the CETA tribunals may decide whether such a measure is 'legitimate' or 'manifestly excessive'. 87 Perhaps the CJEU's conclusion regarding the carve-out in the CETA tribunals' jurisdiction may be best understood as a constitutional message: as long as the CETA tribunals refrain from calling into question the level of EU public interest protection, it will not internally challenge international law through the narrative of primacy of EU law. If read that way, Opinion 1/17 could be characterised as the CJEU's very own Solange decision. 88

Practical consequences for the EU's and the Member States' external investment relations
Opinion 1/17 gave the green light to EU investment dispute resolution in general. The Opinion comes close to operationalising autonomy for IIAs: by carefully pinpointing the procedural and material 'autonomy safeguards' enshrined in CETA, the CJEU established a minimum level of protection for designing IIAs. CETA has thus become the blueprint for the EU's external investment (reform) agenda, as future EU IIAs -including the envisaged MIC 89 -cannot fall behind the standards set in CETA. 90 Moreover, as further argued below, some of the existing IIAs made by the EU and/or the Member States may have to be renegotiated in light of Achmea and Opinion 1/17. The practical implications of these decisions on the EU's (Section 5.1) and the Member States' (Section 5.2) external investment relations can therefore not be overstated.

Implications of Opinion 1/17 for the design of EU IIAs
Opinion 1/17 suggests that in order to be compatible with the autonomy of EU law, the EU IIAs must, at the very least, replicate the standards of protection enshrined in CETA.
Opinion 1/17 gave clear indications as to how the jurisdiction of the CJEU and the ICS can be separated. First, it verified the compatibility of the rule of proceduralisation with EU primary law. That verification might also be relevant outside the confines of investment, such as for the renegotiations of the EU agreement on accession to the ECHR. 91 Second, Opinion 1/17 indicates that EU law cannot form part of the law interpreted and applied by investment courts and tribunals. CETA's 'applicable law' clause is therefore perhaps the most valuable guide for the design of future EU IIAs -including for the MIC. Indeed, we can see that the EU used CETA's ICS as a blueprint for its ensuing or planned IIAs. The stand-alone Investment Protection Agreements (IPAs) with Singapore and Vietnam, 92 as well as the agreement in principle and the textual proposal, respectively, for the modernisation of the Association Agreements with Mexico and Chile, 93 almost literally transcribe CETA's 'rule of proceduralisation' and its 'applicable law clause'. Presumably, the Commission will seek to replicate these two clauses in the multilateral context for the MIC, too. 94 Opinion 1/17 also gave a precise indication as to how EU IIAs may protect the level of public interests desired by the EU legislature. The Opinion suggests that EU IIAs must include at the very least a 'right to regulate' clause, a 'general exceptions' clause, an exhaustive list of cases where 'fair and equitable treatment' may be accorded, as well as a clause defining (indirect) expropriation and generally excluding measures preserving public interest objectives from constituting indirect expropriations. 95 The recently finalised IIAs of the EU include these standards of protection. 96 5.2. Implications of Achmea and Opinion 1/17 for the Member States' BITs and the Energy Charter Treaty The CJEU's recent decisions on autonomy and investment arbitration have legal implications for the Member States' intra-and extra-EU BITs. The Achmea case marked the end of the era of intra-EU BITs. Although Achmea only concerned the ISDS mechanism, and not the BIT as such, the Member States had little interest in maintaining investment protection standards without enforcement. In May 2020, the majority of the Member States concerned agreed to terminate their intra-EU BITs through a plurilateral agreement. 97 So far, the CJEU has not rendered a decision that directly relates to the Member States' extra-EU BITs. Yet, there appears to be no reason why the 'autonomy safeguards' that emerged from Opinion 1/17 should not apply to extra-EU BITs. 98 If extra-EU BITs contain clauses expressly designating national law as part of the applicable law, EU law becomes part of the jurisdiction of ISDS tribunals. By inference, if extra-EU BITs do not comply with the 'material autonomy safeguards', they would equally effectuate that 'a Member State in the course of implementing EU law ... has to amend or withdraw legislation because of an assessment made by a tribunal standing outside the EU judicial system of the level of protection of a public interest established'. 99 The Commission has, in its recent decisions under Regulation No 1219/2012 ('grandfathering regulation'), 100 requested the Member States to separate EU law from the applicable investment law and to include safeguards of their right to regulate. 101  Lastly, the Energy Charter Treaty (ECT) has to be renegotiated in light of Opinion 1/17 and Achmea. The ECT has 53 signatories and contracting parties, including both the EU and the Member States in their own right (multilateral mixed agreement). It contains, among other, an investment dispute resolution mechanism for conflicts between investors and states. The ECT does not include a disconnection clause 102 and could therefore -like intra-EU BITs -apply to conflicts between an investor of one Member State and another Member State. In light of the CJEU's conclusions in Achmea, the Member State inter se application of the ECT is likely to be incompatible with EU primary law. 103 Moreover, the ECT falls behind the standards of protection set in CETA. The Commission has indeed reacted quickly and issued a new statement pursuant to Article 26(3)(b)(ii) of the ECT, which ensures that the EU determines who shall act as respondent in accordance with Regulation (EU) No 912/2014, and underlines the CJEU's competence to 'to examine any question relating to the application and interpretation of [EU law]'. 104 Moreover, the Council adopted negotiating directives for a substantive modernisation of the ECT. 105 In view of the difficulty of reaching consent in a multilateral context -and the danger of watering down standards of protection in order to find compromise -it is uncertain whether the envisaged amendments of the ECT will hold out against the CJEU's novel 'autonomy safeguards'. Either way, further reform will arguably be required to address the Member State inter se application of the ECT. 106 Otherwise, the ECT might not withstand challenge before the CJEU.

Conclusion
In view of its existential importance for the EU's constitutional life on the one hand, and its fuzzy and open-ended conception on the other, it is not surprising that the principle of autonomy raises many complex questions, or that these questions can often only be authoritatively answered by the CJEU. In recent years, autonomy-related questions have arisen with regard to both intra-EU BITs (Achmea) and EU IIAs (Opinion 1/17). In Achmea the CJEU used a high protective threshold for international investment tribunals assessing EU law, which appears to result from the inherent adverse effects of the parallel dispute settlement system established by intra-EU BITs on the internal trust in, and uniform interpretation of, EU law. In Opinion 1/17 the CJEU confirmed that the concept of mutual trust is not applicable in EU external relations and that the reciprocal nature of EU agreements generally justifies the establishment of international dispute settlement mechanisms. This may explain why EU IIAs, in contrast to intra-EU BITs, can preserve the autonomy of EU law -provided that they are designed carefully.
In contrast to intra-EU BITs, CETA contains specific procedural and material 'autonomy safeguards' that essentially seek to separate EU and international investment law, including the jurisdiction of the CETA tribunals and the CJEU. In Opinion 1/17, the CJEU deemed these 'autonomy safeguards' sufficient and declared CETA's ICS compatible with EU law. In contrast to prior case law on autonomy, Opinion 1/17 therefore illustrates precisely how the autonomy of the EU legal order can be protected in a given field (here, investment) through careful treaty design. In other words, Opinion 1/17 operationalised autonomy for EU IIAs. In practice, the CJEU thus ensured that the EU's and the Member States' IIAs will not fall behind the standards set in CETA, which may require redesigning (some of) the Member States' external BITs and the ECT. The need to redesign (some of) the Member States' external BITs might give further impetus to grandfather these agreements with EU IPAs that take due account of the 'minimum autonomy safeguards' set by Opinion 1/17.
The outcome of Opinion 1/17 is thus essentially pragmatic: it enables the Commission to drive forward the EU's bilateral investment agenda and the MIC project, as long as it upholds the CJEU's novel 'autonomy safeguards'. This Solange logic might illustrate the CJEU's self-perception as a constitutional court of a maturing legal order. In this regard, Opinion 1/17 might also be viewed as a rejoinder to the critique against the CJEU's self-referential construction of autonomy that emerged in the aftermath of Opinion 2/13.