The Constitutional Status of EU Agreements: Revisiting the Foundational Questions

II
The Constitutional Status of EU Agreements: Revisiting the Foundational Questions

1. Introduction

This chapter revisits several of the foundational questions pertaining to the constitutional status and legal effects of EU Agreements. These questions are, first, whether EU Agreements upon their entry into force become in principle part of the EU legal order; that is, does the EU adhere to a model of automatic treaty incorporation. A sub-question that flows from an affirmative answer to that question is whether those external treaty commitments are then subject to the lex posterior rule. The second critical question is whether there is scope to constitutionally review the EU’s actual treaty commitments. The third question concerns the test to which EU Agreements are subject in order to have their most potent legal effects, namely, forming a criterion for judicial review.

Constitutional systems may have a text which itself provides answers to some of these questions but it will ultimately fall upon domestic judicial actors to provide the authoritative response. The EU context is no exception and it has been the supreme judicial arbiter within this legal system that has had to answer these core questions. But in the federal-like system that is the EU, these questions acquire additional constitutional salience because as a supranational organization composed of formally sovereign States the answers to these questions have particularly marked constitutional ramifications for its component States. Accordingly, in addressing how the aforementioned questions have been answered in the three sections that follow, this chapter also seeks to articulate the ramifications of the choices adopted for the constitutional orders of the Member States

2. Opting for Automatic Treaty Incorporation: EU Agreements as Acts of the EU Institutions and an Integral Part of EU Law

The ECJ put in place a central plank of the EU’s external relations constitution when it first expressly addressed itself to its jurisdiction over EU Agreements in its Haegeman II ruling that firmly established that the EU was attached to a model that automatically incorporates EU Agreements into its legal order.¹ In Haegeman I a Belgian importer sought to challenge EU measures that allegedly breached the Association Agreement with Greece.² The Court concluded that disputes concerning the levying of charges were to be brought domestically and national courts could then use the preliminary ruling procedure to ensure uniform application of EU law. That importer duly brought proceedings domestically and some 18 months later the Court responded to a Belgian court’s questions as to the interpretation of that Agreement.³ The response centred largely on finding that the relevant charge was a permitted levy under the Agreement and thus not caught by the interdiction on customs duties and charges having equivalent effect. The real significance of the case lay in the assertion of jurisdiction over EU Agreements and the constitutional ramifications to which this gave rise.

Jurisdiction in the preliminary ruling context was provided for in two main contexts, namely interpretation of the Treaty or the validity and interpretation of acts of the institutions. The judicial response to this textual hurdle commenced with an assertion of preliminary rulings jurisdiction concerning the interpretation of acts of the institutions, followed by the assertion that the Agreement was concluded by the Council under Articles 218 and 217 TFEU and was therefore ‘an act of one of the institutions … within the meaning of … Article [267]’. Two more single-sentence paragraphs followed, the first ruling that ‘The provisions of the Agreement, from the coming into force thereof, form an integral part of [EU] law’, and the second affirming preliminary rulings jurisdiction within the framework of this law. The crucial proposition that EU Agreements are both acts of the institutions and form an integral part of EU law have been repeated to this day.⁴ The implications of the stance first taken in Haegeman II for the EU’s external relations constitution and the consequential implications for the external relations constitutions of the EU’s Member States are outlined in the two subsections that follow.

2.1 Automatic treaty incorporation and the EU’s external relations constitution

Haegeman II established that the EU was to be wedded to a model of automatic treaty incorporation. This is evident merely from the unreasoned assertion that once the Agreement comes into force its provisions form an integral part of EU law. The language of being part of domestic law is language expressly written into certain national constitutions in addressing the relationship between legal orders.⁵ The ECJ itself had already used such language in such a context. In the seminal Costa judgment it asserted that the Treaty of Rome ‘became an integral part of the legal systems of the member states … which their courts are bound to apply’. And, thus, if one read Haegeman II alongside Costa, it would appear to follow that whatever is an integral part of EU law is also an integral part of the legal systems of the Member States which their courts are bound to apply. The conclusion flowing inexorably from this, is that such Agreements are in principle capable of possessing those two central distinguishing attributes of EU law: direct effect and supremacy. But the manner in which jurisdiction was established was contestable. The procedure for EU Agreements does require the Council to conclude them and the immediate practice used a Council act to this end. In this sense an act of one of the institutions is at issue. However, it was pointed out that there was a distinction between the EU Agreement and the internal EU act concluding it and that provisions in the former, rather than the latter, were being interpreted;⁶ indeed, it is the Agreement itself that was asserted to be an act of the EU institutions. One response is that in EU practice the Agreement itself is actually included as an annex to the Council act concluding it and therefore part of the act.⁷ Two related points are that EU Agreements are exactly that, Agreements to which the EU is a party and in this sense acts of the EU rather than of an institution and, moreover, they are not unilateral acts, which was arguably the natural remit of Article 267.⁸

There was, arguably, a textually more faithful way to have resolved Haegeman II. Express preliminary rulings jurisdiction exists with respect to the validity of acts of the institutions. Haegeman II was such a challenge,⁹ and therefore jurisdiction could have been asserted simply on this basis. The grounds of validity review are not outlined in Article 267, but if one reads the grounds as coterminous with Article 263,¹⁰ then infringement of the Treaty offers a suitable review ground given that a Treaty provision (Art 216(2) TFEU) expressly provides that EU Agreements bind the institutions.¹¹ Moreover, it had recently been established that non-EU concluded Agreements could be used to challenge EU measures rendering it unsustainable to argue that this would not be possible where EU Agreements were at issue.¹²

The Advocate General had clearly felt more constrained by the text of the preliminary ruling procedure, for he argued that vis-à-vis EU Agreements this jurisdiction only obtained where interpretation was relevant to the validity or interpretation of an act of an EU institution. Had the ECJ followed those promptings it would have generated a very different EU Agreements enforcement model. That approach would have countenanced challenges to EU acts via the preliminary ruling procedure, whilst rejecting cases seeking interpretations where national measures were challenged.¹³ On this account, individuals would have a limited role as enforcers of EU Agreements. The conclusion that the Agreement was an act of one of the institutions assured jurisdiction whenever a national court sought interpretation of an EU Agreement.¹⁴ That reading co-opted individuals and national courts into ensuring compliance with EU Agreements by the Member States rather than just by the EU institutions.¹⁵ It was in embryonic form the external relations counterpart of Van Gend en Loos, for, just as that ruling co-opted individuals and national courts into the enforcement game with respect to Treaty provisions, later extended to secondary measures, so Haegeman II co-opted them with respect to EU Agreements.¹⁶ These constitutional ramifications were somewhat camouflaged because Haegeman II revolved fundamentally around the validity of an EU, rather than national, measure.

It requires little reflection to understand why this construct was pursued, textually contestable though the rationale employed may have been. The EU was beginning to assert itself on the international stage, but was finding resistance in some quarters to its international role,¹⁷ and the judicial reading offered a crucial mechanism for assuring to Contracting Parties that international obligations would be complied with in the EU. Even if this enforcement avenue had not been established, infringement proceedings would still comfortably have accommodated non-compliance with EU Agreements for Article 216(2) TFEU clearly stipulates that the Agreements bind the Member States, thus providing the necessary EU Treaty obligation to police.¹⁸ Challenging EU action via the annulment procedure would also have posed little textual difficulty given the infringement of the Treaty ground of review and that Article 216(2) TFEU asserts that EU Agreements bind the institutions.¹⁹ Thus, for most proceedings jurisdiction over EU Agreements was assured with little textual difficulty, but the role for individuals and national courts that Haegeman II generated was no textual inevitability as the alternative largely-forgotten reading by the Advocate General illustrated. The ‘catch-all’ assertion that provisions of EU Agreements form an integral part of EU law made it crystal clear that all the EU enforcement machinery was available for policing their compliance, even if it was to be some time before this was put into practice. The constitutional significance of Haegeman II should therefore not be underestimated. By assimilating EU Agreements to ‘EU law proper’²⁰ it also ensured that as EU law develops and the remedial tools at its disposal are bolstered, as with the judicial creation of State liability,²¹ they become equally available for the EU Agreements setting.

Article 216(2) TFEU clearly offered a strong textual lynchpin for running arguments for jurisdiction over EU Agreements in most settings. The ECJ, however, made no reference to it despite the presence of precious little constitutional text to rely upon. Arguably, it could have provided a freestanding constitutional anchor for automatic treaty incorporation in the EU legal order.²² The Court was faced with either rejecting decentralized enforcement of EU Agreements vis-à-vis Member States or reconciling that with the text, but it is debatable whether its attempt to do so did less violence to the text than simply basing itself on Article 216(2) TFEU.²³ This could also have been supplemented by Article 19 TEU which required the Court to ensure that in the interpretation and application of the Treaty the law is observed.²⁴

As it turned out, the Court referred to Article 216(2) as rendering EU Agreements binding on the EU and Member States in its very next ruling on an EU Agreement (a case involving domestic action).²⁵ And whilst this axiomatic point was soon repeated,²⁶ it was not until the Kupferberg judgment in 1982 that greater principled consideration was accorded to Article 216(2) and the logic of Haegeman II.²⁷ In Kupferberg Article 216(2) was reiterated with the Court asserting that it was incumbent on both EU institutions and Member States to ensure compliance with EU Agreement obligations. The ECJ then proceeded to reveal why EU Agreements form an integral part of EU law. It was not simply because Agreements were considered to be acts of the institutions as one may have suspected given that this holding directly preceded the single-sentence paragraph in Haegeman II asserting that their provisions were an integral part of EU law,²⁸ rather it was because in ensuring respect for commitments arising from EU Agreements the Member States were fulfilling an obligation above all in relation to the EU which has assumed responsibility for due performance. This ex post justification for Haegeman II was indicating that the reasoning was shaped by the EU’s international responsibility.²⁹ The EU by definition is always a party to EU Agreements, and thus in principle internationally responsible for performance of the obligations assumed thereunder,³⁰ accordingly it would seem logical to underscore the EU law obligation of the Member States.³¹

In federal systems, the centre will usually be equipped with tools to ensure that the sub-units comply with the State’s treaty obligations.³² This, one suspects, did not escape the Treaty drafters given the ease with which a combined reading of Article 258 and Article 216(2) yields such a tool. But Kupferberg involved the preliminary ruling procedure and, accordingly, the international responsibility rationale might be viewed as an additional functional argument in support of general Article 267 jurisdiction. It was curious, however, that the assertion that EU Agreements are an integral part of EU law was first made in a case revolving around an EU measure’s validity and yet eight years later a differently composed bench proffered the ex post rationalization, in a case challenging a domestic measure, that this was so because the Member States have an obligation to the EU. Being an integral part of EU law must surely also relate, to use the federal language, to the centre, rather than just the sub-units, being obliged to comply with its treaty commitments? The judicial explanation could arguably be read more charitably, with the focus being on the EU’s international responsibility and Article 216(2) TFEU rather than merely the obligations of the Member States to the EU flowing from the latter’s international responsibility.

In any event, another critical point drawing on the assimilation of EU Agreements with EU law proper was made:

It follows from the [EU] nature of such provisions that their effect in the [EU] may not be allowed to vary according to whether their application is in practice the responsibility of the [EU] institutions or of the Member States and, in the latter case, according to the effects in the internal legal order of each Member State which the law of that State assigns to international agreements.

The Court then asserted its responsibility via its interpretative jurisdiction for ensuring the uniform application of such provisions throughout the EU. In short, what had been implicit in Haegeman II was confirmed: within the EU legal order the ECJ was to be the authoritative interpreter of EU Agreements.³³

Chapter I noted that while the controversy between the monist and dualist schools was particularly pronounced in the earlier parts of the twentieth century, even into the twenty-first century pieces from prominent scholars have contributed directly to this debate.³⁴ Moreover, within the founding Member States international lawyers had been engaging in heated debate as to how best to conceptualize the legal status and potential domestic effects of treaties.³⁵ Unsurprisingly the monist and dualist conceptual lenses and the debates to which they gave rise were directly transposed to the debate on the internal effect of the EU’s external commitments. As early as 1960, the use of Regulations for concluding EU Agreements had been called for to ensure their internal legal effect.³⁶ Two factors pertaining to Regulations are particularly noteworthy in this respect. The Treaty text states that they are directly applicable and expressly required their publication.³⁷ In contrast, no such publication requirement was imposed for EU Agreements.³⁸ This was no small issue for, as noted in Chapter I, States wedded to a model of automatic treaty incorporation usually stipulate a publication requirement for treaties which if not met has repercussions for their internal legal effects. The Council did not (at least initially) follow the promptings of some commentators in that it opted to use Decisions for the concluding act. However, it would also publish that Decision with the annexed Agreement in the Official Journal thus neutering a potential publication-based objection to the internal effect of an Agreement.³⁹ By the late 1960s the Council appeared to have a change of heart for it commenced with increasing frequency to use Regulations.⁴⁰ One well-placed observer suggested this was out of caution to prevent even inveterate dualists from objecting that the Agreements were not capable of application within the EU.⁴¹ This alteration in practice only fuelled debate as to whether this internal EU measure could be seen as ‘transforming’ the EU Agreements.⁴² This seemingly arcane debate was not devoid of considerable practical consequence. As noted in Chapter I, in Germany and Italy such reasoning had a prominent hold resulting in treaties obtaining the internal rank of the law of approval and therefore the lex posterior rule applied.

The Greek Association Agreement at issue in Haegeman II was concluded via a Decision and the fact that no express significance was attributed to the type of concluding act appeared to resolve any debate as to the need for Regulations to ensure particular internal effects for EU Agreements. Put simply, the assertion that EU Agreements were acts of the institutions and an integral part of EU law owed nothing to the type of internal concluding EU measure. To this day, any attempt to gain purchase out of the type of concluding measure has failed.⁴³ In this sense, the stance first employed in Haegeman II favoured the growing chorus of voices rejecting any notion of the EU concluding act ‘transforming’ EU Agreements.⁴⁴ The apparent need for a concluding act was, however, viewed by some as evidence of dualism insofar as it suggested that the internal EU measure made the Agreement operative within the EU,⁴⁵ but generally Haegeman II has been viewed as monistic by those inclined to use such terminology.⁴⁶ The language of being part of the legal order is associated with legal systems usually characterized as monist in that treaties are automatically incorporated without ad hoc domestic implementation being necessary. Furthermore, Haegeman II could be viewed as implicitly rejecting the lex posterior rule,⁴⁷ in effect the most practical manifestation of the transformation debate at the heart of the monist-dualist controversy. The interpretative engagement with an EU Agreement invoked in a challenge to a later-in-time EU measure unsurprisingly led some to cite Haegeman II as authority for the primacy of EU Agreements over EU secondary legislation.⁴⁸ In any event, many viewed the question as having been answered vis-à-vis international law in general in the International Fruit ruling,⁴⁹ which rendered an alternative outcome concerning EU Agreements untenable, for they, unlike other forms of international law, benefit from a potential constitutional anchor for primacy provided by Article 216(2). Many a commentator has seen in this provision a rejection of the lex posterior rule.⁵⁰ But one can also understand the view of those who do not see it as answering the question.⁵¹ A textually more explicit provision was certainly possible.⁵² This might explain why it was not invoked when the primacy of EU Agreements over secondary legislation was first expressly stated judicially in 1996.⁵³ Indeed, no judicial justification has been forthcoming. One early argument was that it would be inappropriate to apply different reasoning depending on whether it was the relationship between EU law and domestic legal systems or that between EU law and international law.⁵⁴ But as one commentator astutely remarked, if this were so no federal State which stipulates federal law’s primacy over state law would be entitled to apply the lex posterior rule in relation to treaty law.⁵⁵ Moreover, followed through to its logical conclusion, application of equivalent reasoning should require the EU’s international obligations to trump its own constitutional text, precisely the outcome the ECJ demands of the domestic legal order, and precisely the outcome, as Section 3 illustrates, that the ECJ has been unwilling to countenance.

When International Fruit was decided, the Member States adhering to the lex posterior rule were outnumbered by four to two, but by the Haegeman II ruling this had become five to four the other way. By the time of the first express acknowledgement of the hierarchically superior status vis-à-vis secondary EU law in 1996, three more States had joined each group and thus eight of 15 States were domestic adherents to the lex posterior rule.⁵⁶ This divergence in domestic approach offers a strong argument in favour of rejecting the lex posterior rule to the benefit of all EU Member States. To have adopted the lex posterior rule would have resulted in four of the founding States being able to achieve collective outcomes through their EU treaty-making power combined with their EU legislative output that would not be possible with respect to their non-EU treaty-making and law-making capacities; later-in-time EU secondary legislation could trump EU Agreements but later-in-time Member State legislation could not trump that Member State’s own treaty commitments. The corollary to this is that rejection of the lex posterior rule results in an outcome not ordinarily possible (outside the field of EU law) for adherents to the lex posterior rule; their collective exercise of the EU treaty-making power could not be trumped via their collective exercise of the EU legislative process, but precisely this outcome could be achieved where the domestic treaty and law-making processes were employed. The logic for the lex posterior rule is usually traced back to this being a later expression of the democratic will, but if the German or Italian model were transposed to the EU level this would not result in the Italian or German legislative will being able to trump such Agreements. Rather, the EU’s legislative output would theoretically be capable of trumping such Agreements.⁵⁷ As qualified majority voting applies, that would mean that States such as Germany or Italy could find EU Agreements being trumped by EU legislative output for which they were outvoted in the Council. In sum, from the perspective of both the Member States adhering to the lex posterior rule and those that did not, the case for EU Agreements to have a hierarchically superior status internally than EU legislative output is compelling.

2.2 Automatic treaty incorporation and the external relations constitutions of the Member States

The ramifications of the Haegeman II ruling, as further articulated in Kupferberg, were substantial for the external relations constitutions of the Member States. This was palpably so for those States attached to non-automatic treaty incorporation. For the three Member States wedded to such an approach at the time of these early judgments (Denmark, Ireland, and the UK) the effect was to convert them into automatic treaty incorporation States for a category of treaty law. One category of treaties, to which they may or may not themselves be parties, would through the conduit of EU law become automatically incorporated into their domestic legal orders,⁵⁸ sitting alongside the other categories of treaty to which they are a party but that do not, in principle, subject to ad hoc domestic implementation, become part of the domestic legal order. This accordingly qualifies the classification in Chapter I of Denmark, Ireland, and the UK as non-automatic incorporation States, for the constitutional effect of EU membership is that there is a growing category of treaties for which their orthodox constitutional model no longer applies. In such States, the domestic legislature no longer determines the role, if any, that domestic courts play with respect to this particular category of treaties. Rather, the ECJ assumed that role as Haegeman II intimated and as Kupferberg made explicit. This is unquestionably of momentous constitutional import for those States and any other State wedded to the non-automatic incorporation model that has since contemplated EU accession (such as the Nordic group that joined in 1995).

Whilst the impact of automatic treaty incorporation for EU Agreements was thus most profound for the non-automatic treaty incorporation States, one should not underestimate the impact that the emerging construct would have on the States already familiar with automatic treaty incorporation.⁵⁹ Most obviously, it would no longer be for their courts alone to determine the legal effect and interpretation of this category of treaty, rather this became a role in which they were supervised by the higher authority of the ECJ; this meant that the direct effect determination, the traditional preserve of the national court when dealing with treaties,⁶⁰ would not be their authoritative prerogative for EU Agreements. This is of critical significance for whilst national courts have on occasion been guilty of shielding their domestic legal order from the impact of treaties, there would be an important category of Agreements for which they would no longer have free rein. And endowing EU Agreements with the general attributes of EU law logically included a hierarchically superior status to domestic law. The later-in-time rule which remains the default rule in Germany and Italy gives way not only to EU law proper but would also, in principle, now do so with respect to EU Agreements. Indeed, the logical implication of this assimilation to EU law proper is that their status became superior to that of the domestic constitution itself and only one of the founders—the Netherlands—accorded treaties a supra-constitutional status.⁶¹

Furthermore, whilst legislatures in automatic treaty incorporation States would have a constitutionally entrenched role in assenting to at least certain types of treaty they would no longer have such a role for pure EU Agreements.⁶² Indeed, here the initial EU parallel with domestic automatic treaty incorporation models broke down: the European Parliament’s input into the treaty-making process was astonishingly meagre until the Maastricht Treaty gave it an assent requirement for some categories of significant Agreements,⁶³ but it was not until the Lisbon Treaty that an assent requirement became the norm (Art 218(6) TFEU). The EU model, therefore, vastly empowered the domestic executive in external relations via its control of the EU’s treaty-making powers, the significance of which was accentuated by the exalted legal status of this output. EU Agreements thus exacerbated the impact of EU membership on the domestic separation of powers: it increased the empowerment of domestic executives via their new collective treaty-making role, with a concomitant disempowerment of the domestic legislature, and an empowerment of the domestic and supranational judiciary for whom another category of EU law emerged.

The constitutional ramifications of adherence to a fully-fledged automatic treaty incorporation model signified by Haegeman II can be further illustrated when viewed alongside the substantial expansion of the EU’s limited express treaty-making competence via the implied treaty-making powers doctrine. The effect of the seminal ERTA jurisprudence was that over time as internal EU legislative output expanded, there was a greater corpus of law capable of being affected by an independent treaty-making power of the Member States and for which the EU instead would accordingly acquire the treaty-making power in the affected sphere. This reduction in the Member States’ independent treaty-making power was replaced by the capacity to assume contractual obligations with third parties via the EU framework, only now the internal legal effect of such norms was not to be authoritatively determined by their domestic constitutional order but rather by a supranational court.⁶⁴ Nor could the EU legislative process itself be used domestically to trump these external obligations;⁶⁵ appropriate as this outcome may be, as suggested above, it does nothing to detract from the domestic constitutional ramifications for Member States as compared to the position that would exist in the EU’s absence.

The constitutionally powerful repercussions for the Member States of the EU Agreements model to which the Treaty text combined with judicial rulings gave rise, was accentuated by two predictable developments. First, the emergence of qualified majority voting for the treaty-making process, which post-Lisbon has become the norm, such that the constitutional impact of EU Agreements for any specific domestic legal order can apply even though that State’s Council representative voted against its conclusion. Secondly, the rapid expansion of the EU’s treaty-making activity thus forming an increasingly significant component of the treaty obligations to which Member States are bound.⁶⁶