Learning from failure in ‘an Integral Part’ of EU Law: interpretation of international treaties in the CJEU

Abstract

The Court of Justice of the European Union has been criticised increasingly for its approach to international law. While much literature focuses on reluctance to apply international law (ie refusing direct effect), this criticism also includes interpretation, arguably a more contentious area. The Court interprets international treaties through either the Vienna Convention or its own teleological method, with the latter increasingly applied. The legitimacy of applying localised methods of interpretation to international treaty law is debated in scholarship. Whether the Court’s case law applying international law is Völkerrechtsfreundlichkeit (friendliness to international law) is the focus of most contributions in this area. This Article seeks to move beyond that debate by demonstrating the Court should in all instances seek to achieve the legal imperatives of certainty and justice. It is explained that justice divides into ‘thin’ and ‘thick’ forms. Whilst ‘thin’ justice is widely accepted and amounts to treating like cases alike, substantive (‘thick’ justice) outcomes are inherently debatable. The Article proves the Court is failing to clearly distinguish cases (‘thin’ justice) and that case law is uncertain. There are also significant questions concerning ‘thick’ justice. The Court has been subject to criticism for substantive outcomes in this area, with the Commission and Council even seeking to limit its role. With case law that is uncertain and appears unjust it is argued that there is failure in this ‘integral part of EU law.’ The Court is now under increased pressure, and it is uncertain how it will respond. There are certainly cautionary lessons to be learned concerning the importance of paying proper attention to justice and certainty for EU law as a whole, and beyond.

1. Introduction

The Court’s varied approaches to interpreting international law will be measured in this Article against the imperatives of justice and certainty. Legal certainty (providing anticipatable rules to which conduct will be subjected) and justice (achieving the right outcome in an individual case) constitute the dual aims of law. ¹ It is frequent to debate whether the EU courts are sufficiently respectful, open or friendly towards international law. However, ‘Völkerrechtsfreundlichkeit ² is a matter of perspective: while the international lawyer may claim the glass is half empty, the EU lawyer may suggest it is half full’. ³ Instead, applying certainty and justice provides greater objectivity. It is hoped that by moving beyond purely a debate over friendliness towards international law, clearer findings concerning the Court’s case law can be reached. ⁴ The Article will expose failures in this area of law and provide concrete lessons of broader relevance for EU law as a whole, and beyond.

2008 featured three damaging cases concerning the EU’s respect for international law ⁵ and since then criticism of the Court’s case law applying international law has become widespread, even appearing in the Wall Street Journal. ⁶ Prominent cases drawing criticism have arguably focused on application of international law and related EU concepts of autonomy and direct effect. There, however, an ‘elementary divide’ ⁷ leaves domestic courts (including, many suggest, EU courts ⁸ ) free to decide whether to apply international law or not. Interpretation gives rise to many different issues and remains a more greatly contested sphere than that of direct effect. In spite of receiving less attention in EU external relations scholarship, interpretation of international law clearly plays a decisive role in many cases; without interpretation fleshing out the relevant rights relied upon ‘the finding of direct effect is fruitless’. ⁹

A wealth of literature covers the topic of direct effect and it is not proposed to add significantly to these contributions. ¹⁰ More recent developments aimed at restricting the Court’s role (including through refusing direct effect) are, however, included. ¹¹ There is also a significant yin-yang relationship between direct effect and interpretation; often, if direct effect readily allows reliance on international law, then interpretation will become more flexible to enable the Court to retain discretion in deciding case outcomes. ¹² Accordingly, whilst the focus of this Article is on the role of interpretation in achieving the balance between justice and certainty, we will necessarily be cognisant of interpretation’s relationship with direct effect and other rules controlling reliance on international law throughout. ¹³

For interpretation two different options have emerged for the Court in this area of law; the Vienna Convention on the Law of Treaties (VCLT) and teleology.

International treaties have widely accepted criteria for interpretation contained in the VCLT. ¹⁴ Attempts to follow VCLT ensure interpretation remains anchored in international practice. However, the requirements of VCLT interpretation and the extent to which they are construed as strict varies in scholarship. ¹⁵

Concerning EU law, since the early 1960s the Court eschewed more conservative interpretative approaches (which some argue are reflected in VCLT) and instead adopts heavily purposive – teleological – interpretative approaches to EU law, which is itself a product of international treaties.

There is a stark divergence between VCLT and teleological interpretation, at least as understood by the Court itself. ¹⁶ The paths are not just divergent in method but also in destination, often resulting in very different outcomes. ¹⁷ As such, knowing when the Court will select which method is important. Whereas teleology is always applied internally, in external cases the picture is more mixed.

For example, international treaties concluded by the EU are ‘an integral part’ of EU law, which the Court has found it has the power to interpret. ¹⁸ In the face of a somewhat hybrid internal/external definition of where international treaties sit in the EU legal order, the Court has opted not to conclusively pursue one of either VCLT or teleological interpretation. The Court has acknowledged VCLT’s relevance for interpreting international treaties and shown some willingness to use it, but not exclusively. Bilateral treaties to which the EU is party, particularly, although not only, where such treaties anticipate(d) future accession to the EU by the partner State have been subject to teleology instead of VCLT. ¹⁹ Teleological interpretation has often and increasingly been applied to bilateral agreements, although more recent interventions by other EU institutions seek to stall this progression. ²⁰ Multilateral treaties to which the EU is party have also seen teleological interpretations, including where the Court purports to be applying international rules. The latter manifests itself as a mutated version of VCLT. ²¹ This practice appeared to have paused concerning multilateral treaties, but it has re-emerged recently. ²² Opportunistic interpretations of international law are also especially prevalent in the Court’s case law concerning treaties which may impact on third parties. ²³

The case law which results is very problematic; there is significant uncertainty, and it also appears unjust. This means the two fundamental aims of adjudication seem not to be met. The stumbling block which stops us tantalisingly short of concluding that the case law is unjust is attributable to the necessary division of justice into its ‘thin’ (treating like cases alike and unlike cases unlike to the extent of their unlikeness) and ‘thick’ (focusing on the substantive outcome) forms. ²⁴ It will be seen that in the Court’s case law the rationale for distinguishing cases is unclear, evidencing shortcomings in ‘thin’ justice. Given that ‘thin’ justice is problematic and there is significant uncertainty in the case law, what is left for the Court is a final Hail Mary pass ²⁵ – recourse to substantive justice alone. The actual outcome in an individual case is the manifestation of competing aims or norms (eg environmental protection, trade interests or fundamental rights) as applied to specific facts. ²⁶ Appealing to substantive justice to legitimate a case is essentially the claim that the specific decision was a ‘good’ one or the ‘right’ thing to do. This is highly subjective. There has been criticism of case law from scholarship, and other EU institutions have even sought to limit the Court’s role or capacity to interpret widely. ²⁷ In practice, many outcomes arrived at by the Court have been contentious. Oscillations in the case law also imply that the Court may lack a consistent overarching vision of substantive justice. Due to substantive justice’s contested nature, we cannot stop the Court throwing its Hail Mary pass, even though we may be very sceptical.

Ultimately, subject to the narrow caveat concerning substantive justice, this area of law is uncertain and unjust, marking it out as failed.

The Article covers the following: first, justice and certainty are introduced and developed; second, key differences between VCLT and teleological methods of interpretation are explained; third, the Court’s interpretative approaches to bilateral and multilateral treaties, and treaties impacting third parties are analysed (this is undertaken in three parts with a loose thematic division applied to support analysis, the first placing greater emphasis on ‘thin’ justice, the second on ‘thick’ justice and the third on legal certainty). A conclusion identifies where improvements can be made, especially concerning legal certainty within individual cases. Overall, though, the wider picture is one of ongoing tumult, with new protagonists, uncertain dynamics and less than clear objectives. Ultimately, the study provides a cautionary reference point. It serves as a reminder of the risks for all courts in neglecting justice and certainty.

2. Justice and legal certainty

A. Preferring justice or certainty

Established approaches to interpretation serve to provide greater certainty as to how a case will be resolved through repetition of recognised methods, especially if the method is conservative in nature and the rule it is applied to is clearly formulated. ²⁸ However, legal certainty does not occupy the whole field when we are considering that which is desirable in adjudication. Justice (achieving the right outcome in an individual case) is the second aim. There is a frequent tension within adjudication concerning the appropriate balance between legal certainty and justice. Twining describes it as ‘[t]he most persistent of all problems for the [person] of law’ to reconcile these demands. ²⁹ Other scholars have variously identified the ‘dual nature of law’; ³⁰ that ‘the law should strive to balance certainty and reliability against flexibility’; ³¹ that these are the ‘two conflicting and yet equally important objectives in any legal order’; ³² the need ‘to guarantee simultaneously the certainty of law and its rightness’; ³³ for ‘[t]he law must be certain. Yes, as certain as may be. But it must be just too.’ ³⁴

The question then becomes; when ought we to depart from legal certainty in pursuit of justice, or accept an impact on justice to provide legal certainty? Scholarship is divided on this.

Dworkin conceives of ‘law as integrity’ through imagining a superhuman judge, Hercules, who would ‘test his interpretation of any part of the great political structure and decisions of his community by asking whether it could form part of a coherent theory justifying the network as a whole.’ ³⁵ As such, whilst seeking to unify justice and certainty in a ‘right answer thesis’, Dworkin can ultimately be considered to place a greater emphasis on justice than legal certainty. After all, a superhuman judge has one obvious flaw; their impact on legal certainty for us mortals. ³⁶

In stark contrast, Weber advocates the merits of ‘formal rational law’ (through legislation) supported by ‘formal rational administration of justice’. ³⁷ This means conserving a high degree of legal certainty. Weber suggests that the ‘modern judge is like a vending machine into which the pleadings are inserted together with the fee and which then disgorges the judgment with its reasons mechanically derived from the Code.’ ³⁸

We could add other academics in between. Radbruch, having initially prioritised certainty even more greatly, ³⁹ subsequently suggested that if ‘conflict between statute and justice reaches such an intolerable degree’ then ‘the statute, as “flawed law,” must yield to justice.’ ⁴⁰ Alexy aligns himself closely with Radbruch’s model, ⁴¹ whilst Cardozo cites Munroe Smith with approval. Smith suggests that ⁴² :

[A] result which is felt to be unjust…may not be modified at once, for the attempt to do absolute justice in every case would make the development and maintenance of general rules impossible; but if a rule continues to work injustice, it will eventually be reformulated.

Thus, unlike Radbruch, it is not the weight of injustice in a particular case which demands change, but persistent injustice.

Accordingly, there is significant divergence, and a range of possible options emerge. Others, for this reason, decline to take a position. Llewellyn suggests that an ideal model cannot be achieved unless ‘an arbitrary choice is made between the elements of certainty and justice’. ⁴³ Rawls is also notable in his hesitancy on the matter. Identifying first that ‘[e]ven where laws and institutions are unjust, it is often better that they should be consistently applied. In this way those subject to them at least know what is demanded and can try to protect themselves accordingly’. ⁴⁴ He subsequently adds, ‘[o]n the other hand, it might be better in particular cases to alleviate the plight of those unfairly treated by departures from existing norms.’ ⁴⁵ Rawls is essentially identifying that this may be a very context-specific question, and ultimately labels it ‘one of the tangled questions of political science’. ⁴⁶

It is submitted that we need not go further to identify that there is not agreement in abstract on an appropriate balance between justice and certainty: vending machines and Hercules have little in common. Having identified that we cannot deploy a unified imperative of justice and certainty we will now engage with the component elements of these concepts and consider their relationship with one another. Before we do so we will briefly highlight the continuation of the justice and certainty problem in international interpretation, especially when related to domestic courts.

B. Replication of the justice and certainty problem in International Law’s interpretative requirements for ‘Domestic Courts’ problem

It has been highlighted above that debate concerning the CJEU’s application and interpretation of international law is frequently focused on whether it is sufficiently friendly towards international law itself. Below we will explore in detail the key elements and distinctions between the VCLT and the Court’s own teleological approach to interpretation, ⁴⁷ but here we can note some of the main challenges in holding up international law and its interpretative standards as a measure of the case law. Prior to doing so it is important to give a brief explanatory note on this Article’s treatment of EU courts as ‘domestic’.

There is undoubtedly awkwardness that arises concerning EU courts compared to ‘other’ domestic courts, not least because the Court initially slipstreamed international law’s legitimacy to develop its own. ⁴⁸ Unique history and continued attempts of the EU to project its respect for international law are relevant factors in considering the relationship between EU and international law. ⁴⁹ For these reasons any hostility towards international law often feels less acceptable coming from EU courts. ⁵⁰ But whether this justifies refusing the EU courts’ admittance to the category of ‘domestic’ is more questionable.

Scholars have identified variously that the Court considers itself to be a municipal court, ⁵¹ that whilst ‘a judicial organ of a regional organisation established by virtue of international law …. it has become more and more common, however, to regard the ECJ as being functionally equivalent to a municipal court’ ⁵² and, more robustly, that it is ‘a municipal court, a court of the European Union, not a general international court’. ⁵³ Von Bogdandy’s reasons for treatment of the Court as ‘domestic’ are especially persuasive: ⁵⁴

…the EU is based on the principle of vertical and horizontal constitutional compatibility; and given its essentially unitary political system, which is rooted in its territory and citizens; its judiciary, which is endowed with strong competences; and its largely parliamentary legislative. All this – in short, a federal unity – cannot be found outside the Union.

We could add to the list that to refuse the EU courts’ domestic status would also create a deficit in rights for Member States compared to other States, where the latter are free to control domestic effects of international law. Member States would see sovereign rights disappear into the ether.

In treating the Court as ‘domestic’, then, it is important to consider how international interpretative obligations might apply to it. VCLT is debated concerning the clarity and strictness of interpretative obligations it imposes on courts. Even at ‘international’ level there is divergence between more textual approaches (eg of the WTO and ICJ) and more expansive approaches (eg of ECtHR and – where included – the EU itself). ⁵⁵ Some argue that these divergences can be accommodated through VCLT’s elasticity, whereas others take the view that certain regimes develop bespoke methods beyond VCLT. ⁵⁶ This debate, increasingly, is not limited to specialised courts in specific regimes, it attaches instead to types of treaties, for example human rights treaties. ⁵⁷

The consequences of the debate are not always entirely clear. Even assuming agreement could be reached that certain regimes or individual treaty interpretations fall outside of VCLT methods, this does not end debate over the acceptability of such approaches. Indeed, to the extent that the ICJ’s approach to human rights protection may fall outside of VCLT (and contrast with its own practice), is this problematic or desirable? There is also increased tendency to question the tools themselves rather than the use made of them; Crawford and Keene note debate whether VCLT is ‘sufficient or even applicable in the interpretation of human rights treaties’. ⁵⁸ Of course, human rights may be noted as a case apart, but open questioning of VCLT’s relevance has emerged in other areas too, such as investment treaties. ⁵⁹

The specific context of domestic courts also adds further complexity to interpretative rules that are less than hard and fast. For some scholars the acceptance of the interpretative community is characteristic of whether an interpretation is ‘good’. ⁶⁰ When dealing with international law the scope of the relevant community to be considered in this regard broadens, ⁶¹ therefore Halberstam observes ⁶² :

…the interpretation of international law is still ultimately a collective endeavour, ie a shared enterprise among the various judicial and other participants around the world who lay claim to interpret these common legal norms. Domestic courts must accordingly not give a purely partial interpretation that considers only their pluralistic point of view…

But for others ⁶³ :

It would be simplistic to limit the role of national courts in the international legal order to that of ‘enforcers’ of the law. By interpreting and applying international obligations, national courts may contribute to their development.

This hints at the fact that domestic courts also form part of a State, for whom ‘[e]ach deviation [from international law] contains the seeds of a new rule.’ ⁶⁴ Such practice even takes on artistic aesthetics for some, ‘like musical players in an orchestra, these players need not necessarily play the same tune with the same musical instrument.’ ⁶⁵

There is at least some limited convergence in scholars’ views: those who advocate flexibility often do not conceive this as unlimited (that would risk ‘conveying the impression that “anything goes” and that it is hence not law after all’ ⁶⁶ ), equally, those seeking to impose greater rigour concede that inevitably some discretion remains and outcomes can differ. ⁶⁷ Ultimately, though, there (unsurprisingly) remains debate over the precise point at which ‘localised expression’ intersects appropriately with international law’s ‘universalist aspiration’ for interpretation. ⁶⁸

From this perspective asking whether the case law is sufficiently respectful of international law emerges as a less than helpful benchmark from which to draw clear conclusions. International law is also conflicted concerning certainty and justice, not least as concerns domestic courts. Focusing primarily on how friendly towards international law the EU courts are will not be sufficient and can only delay the final analysis. Instead, we will proceed by recognising that there is not a unified model, the Court must instead pursue the two recognised imperatives of adjudication; legal certainty and justice. ⁶⁹ These are the essential elements in adjudication and we will now interrogate them further. It will emerge that they can function as clearer benchmarks against which the case law can be measured.

C. Defining certainty, and ‘Thick’ and ‘Thin’ justice

Without a unified theory of justice and certainty it becomes necessary to consider their individual properties more closely.

Within the EU, the Court has held that ‘[t]he principle of legal certainty is a fundamental principle of Community law which requires, in particular, that rules should be clear and precise, so that individuals may ascertain unequivocally what their rights and obligations are and may take steps accordingly.’ ⁷⁰ AG Sharpston lists ‘guises’ of certainty appearing in case law as ‘stability, unity and consistency’ with the aim ‘to ensure that situations and legal relationships governed by EU law remain foreseeable’. ⁷¹ Of course, in the EU, too, legal certainty is not all that needs to be achieved and the principle is not absolute; it may need to be balanced against other principles linked to justice. ⁷²

Legal certainty is conceptually less challenging than justice and is more easily identified in practice. Even if we disagree on the lengths we ought to go to in achieving legal certainty, ⁷³ it is likely that commonality would be present in our conception of it. Given that legal certainty is recognisable, decisions lacking legal certainty are preliminarily in a higher risk category. This is so because if ultimately uncertain approaches and outcomes do not do (or at least are not seen to do) justice then there is nowhere left to hide. ⁷⁴

However, due to ‘thick’ (or substantive) justice’s contested nature – Nicol suggests we ‘celebrate justice as something we argue about’ ⁷⁵ – individual outcomes arrived at by the Court are not easily proved to be incorrect. Substantive justice in its thickest form could be thought of as being akin to Dworkin’s ‘right answer thesis’; it would encompass both legal certainty and justice, and would give a single correct answer to a given case. ‘Thick’ justice in this Article is imagined as falling short of this ideal and instead is focused on justice credentials separated from legal certainty. It can be thought of as whether this was a ‘good’ thing to do or aim for in a given case. Even without the final additional complexity (or perhaps impossibility) of unifying justice and legal certainty, substantive justice will often be debatable as ‘there exists no common standard, no single measurement for justice’ and ‘it is probably fair that no such theory can be developed.’ ⁷⁶

‘Thin’ (or formal) justice, however, is widely accepted. ‘Thin’ justice ⁷⁷ amounts to the assertion that like cases should be treated alike and unlike cases unlike in proportion to their unlikeness. ⁷⁸ So, for example, we may agree that twin children should (all things being equal) have the same bedtime (‘thin’ justice). But it will inevitably be more debatable as to whether that bedtime should be 7pm, 8pm, 7:19pm etc. (‘thick’ justice).

Hart states that the maxim ‘treat like cases alike’ must remain ‘incomplete’ until we know ‘what differences are relevant’ ⁷⁹ and Lyons states that ‘[t]here are innumerable ways in which cases may be classified as alike or different, since any set of cases share some properties and fail to share others.’ ⁸⁰ Perelman, for these reasons, expresses concern that such analyses will remove us from the sphere of formal justice and return us to (contested) substantive justice. ⁸¹

But we will (at least) have narrowed to a closer point of enquiry, and at that stage Hart suggests that ‘the resemblances and differences … which are relevant for the criticism of legal arrangements as just or unjust are quite obvious.’ ⁸² For example, with our twin children, if one of them was unwell or had a particularly exhausting day that child may be expected go to bed sooner, whereas if one had an afternoon nap that child would not be. The fact that one of them wore a blue sweater and one a red sweater would likely be irrelevant to our decision.

Where differences are slight they will not call for radically different outcomes. ⁸³ In applying ‘thin’ justice, then, we would not expect that every case regardless of subject matter is decided similarly, instead we would expect proportional difference. This separates ‘thin’ justice from being a simple, partial restatement of legal certainty. ‘Thin’ justice also enables us to indirectly question ‘thick’ justice; consistent and coherent adherence to a vision of substantive justice by a court can be expected to lead to formal justice. In our example, if the bedtimes of our twins varied significantly each day for both or either of them without clear reason then that would be a shortcoming in ‘thin’ justice and might imply a flaw in the concept of substantive justice being applied. ⁸⁴

Taking this one step further, in spite of analytical limits concerning substantive justice, pursuit of it at the expense of certainty can ultimately be identified as self-defeating if the newly arrived at conception changes in a subsequent case without external factors (the unlikeness) dictating such. That former case – which departed from legal certainty in pursuit of substantive justice and is now overruled (or moved away from) – has nothing of merit left to it: it is very likely to be uncertain and unjust. ⁸⁵

D. Justice and certainty: unique dynamics

Whilst above we have so far treated certainty and justice (especially substantive justice) as sitting on opposite sides of balancing scales, this is not always so. In particular there is an interesting dynamic by which, first, a greater divergence between justice outcomes intensifies the need for legal certainty and, second, pursuing certainty or justice in a given case does not preclude further steps to improve the other element simultaneously.

First, it is submitted that the greater the divergence between justice outcomes, the more detrimental to legal certainty will be any indeterminacy. For example, punishments for parking violations being (a) warning, (b) parking fine issued, or (c) car is towed and destroyed: the justice difference between (a) and (b) is relatively minor whereas that between (a) and (c) is significant and for this reason warrants greater certainty (eg through signage). ⁸⁶

The Court has (at least) two methods available to interpret international treaty law and given that its deployment of VCLT and teleological methods differ so greatly ⁸⁷ a key step is the decision of which method to apply. In this light any uncertainty exposed in that step is especially problematic.

Second, legal certainty and justice constrain each other, but they do not do so absolutely. In particular, pursuit of justice does not completely limit legal certainty. In this regard the suggestion that ‘departures from past cases must be carefully justified’ is strongly endorsed. ⁸⁸

Keck, an internal EU case concerning free movement of goods, is a good example of how this ideal may begin to manifest itself ⁸⁹ :

In view of the increasing tendency of traders to invoke [now Article 34] of the Treaty as a means of challenging any rules whose effect is to limit their commercial freedom even where such rules are not aimed at products from other Member States, the Court considers it necessary to re-examine and clarify its case-law on this matter.

It went on to say that ‘contrary to what has previously been decided’ in Dassonville ⁹⁰ ‘certain selling arrangements’ would be outside of the scope of the provision ‘so long as those provisions apply to all relevant traders operating within the national territory and so long as they affect [traders] in the same manner, in law and in fact’. ⁹¹

This marked a significant change to justice, but detailed reasoning covering departure from previous rules/case law, the reason for this, and indication of the scope and enduring nature of the decision is provided. There has been plenty of complexity concerning Keck since, ⁹² but as an amelioration of the impact of justice changes on legal certainty the effort at forthrightness is notable.

E. Summary of key findings for justice and certainty

We have identified that there is not a unified model for the appropriate balance between certainty and justice (achieving the right outcome in an individual case) which can be applied. Instead, we will identify how the Court’s case law formulates a balance between the two. It has been suggested that departure from legal certainty is necessarily a risk and places a greater emphasis on the need to achieve justice.

In achieving justice, ‘thin’ justice suggests that we should see proportionate differences between decided cases (treating like cases alike and unlike cases unlike to the extent of their unlikeness).

We have noted that whilst adherence to ‘thin’ justice is more readily identifiable, adherence to substantive (‘thick’) justice will often remain contested. Inconsistency in ‘thin’ justice will, though, make us more sceptical concerning the presence of an overarching vision of substantive justice. It is also appropriate for us to question substantive outcomes of individual cases, especially if these are subsequently recanted by the Court.

Moreover, whilst there is frequently a tension between justice and certainty, it is important to note that pursuit of justice does not operate as a cap on legal certainty in all instances.

Having considered key aspects in the balancing of justice and legal certainty and their merits as benchmarks against which to measure the case law, the interpretative methods which are utilised by the Court can now be engaged with.

3. Teleological and Vienna convention interpretation compared

The teleological method is the Court’s ‘standard’ approach to interpretation of EU law. ⁹³ Indeed, where other approaches – such as literal, schematic or historic ⁹⁴ – do not fit with the teleological approach they are rendered secondary. ⁹⁵ In CILFIT the Court explained what the teleological method requires ⁹⁶ :

[E]very provision of community law must be placed in its context and interpreted in the light of the provisions of community law as a whole, regard being had to the objectives thereof and to its state of evolution at the date on which the provision in question is to be applied.

Thus the ‘teleological’ approach has been noted to eschew a close literal reading in favour of a more purposive approach. ⁹⁷

Some suggest the Court goes further in its approach to interpretation as ‘[i]t is not simply the telos of the rules to be interpreted that matters but also the telos of the legal context in which those rules exist.’ ⁹⁸ The latter is referred to as ‘meta-teleological’ reasoning ⁹⁹ and forms part of what is for convenience here termed the teleological approach, although it goes beyond the CILFIT criteria and instead takes a broader view of what can be considered, further prioritising the Court’s capacity to do justice. This approach provides the Court with a means to reinterpret specific norms based on the consequences of its own previous rulings ¹⁰⁰ and even enables it to bring new norms into existence. ¹⁰¹

For international treaty interpretation the VCLT holds a dominant position. Whilst applicable to States, it expressly refers to the fact that the Convention does not prevent its rules being applied independently as customary international law. ¹⁰² The Court itself has acknowledged that ‘a series of provisions in that convention reflect the rules of customary international law which, as such, are binding upon Community institutions’. ¹⁰³

The most frequently arising provision concerning the Court’s deployment of VCLT is Article 31 – the ‘general rule of interpretation’ – which it is helpful to cite in full:

1. 1. A treaty shall be interpreted in good faith 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.

2. 2. The context for the purpose of the interpretation of a treaty shall comprise, in addition to the text, including its preamble and annexes:

   1. (a) any agreement relating to the treaty which was made between all the parties in connexion with the conclusion of the treaty;

   2. (b) any instrument which was made by one or more parties in connexion with the conclusion of the treaty and accepted by the other parties as an instrument related to the treaty.

3. 3. There shall be taken into account, together with the context:

   1. (a) any subsequent agreement between the parties regarding the interpretation of the treaty or the application of its provisions;

   2. (b) any subsequent practice in the application of the treaty which establishes the agreement of the parties regarding its interpretation;

   3. (c) any relevant rules of international law applicable in the relations between the parties.

4. 4. A special meaning shall be given to a term if it is established that the parties so intended.

It is immediately apparent that application of these rules is not a straightforward matter (together with Article 32, allowing for recourse to supplementary information if an interpretative outcome is unclear or unreasonable, they form the rule). ¹⁰⁴ The elements of Article 31 are suggested not to be hierarchical, but instead a ‘logical progression’ ¹⁰⁵ or ‘progressive encirclement’. ¹⁰⁶ The three elements of interpretation contained within can be seen as textual, contextual and purposive (or teleological). Aust suggests that ‘although paragraph 1 contains both the textual (or literal) and the effectiveness (or teleological) approaches, it gives precedence to the textual.’ ¹⁰⁷ This is arguably supported further by the aim of deriving the intention of the parties to the treaty. ¹⁰⁸

Whilst the rules do not preclude ‘evolutionary’ interpretation of terms, this often remains associated with party intention. ¹⁰⁹ Accordingly, there is provision for localised interpretation where ‘subsequent practice … establishes the agreement of the parties regarding its interpretation.’ ¹¹⁰ Agreement in advance would overlap with Article 31(3)(a), ¹¹¹ but a ‘bona fide’ effort at interpretation is anticipated, ¹¹² which arguably implies a genuine prospect of acceptance by others when making the first move in a new interpretative direction. As such, it is suggested that domestic courts ‘avoid meticulously the temptation of indulging in legal parochialism, universalising concepts that are in fact peculiar to their respective legal systems’. ¹¹³ This, of course, is reinforced by the prominent obligation of Article 31(1) to interpret in ‘good faith’. However, as we have previously noted, ¹¹⁴ VCLT is far from dispositive and provides significant scope for debate.

Some downplay the distinction between EU teleology and VCLT, ¹¹⁵ but key differences often appear between the methods in practice. ¹¹⁶ Klabbers suggests that ‘[w]hile it goes too far to suggest that ‘anything goes’ under VCLT, still, ‘quite a bit goes’’. ¹¹⁷ Beck argues, though, that in spite of VCLT’s flexibility – retained specifically in response to State concerns ¹¹⁸ – the CJEU’s teleological approach still falls outside of the rules. ¹¹⁹ Even if it may be true that the difference is due to misconception – the EU Court and lawyers have been suggested to treat VCLT as ‘a crude method of interpretation confined to the text…that bears little resemblance with international law’ ¹²⁰ – the attitude taken, even if misplaced, carries implications in practice. Therefore, regardless of debate concerning VCLT’s true nature, in the Court’s ‘recurring (though not systematic)’ approach to VCLT, ¹²¹ it formally considers it to require a distinct approach to that of teleology. ¹²² Indeed, it has been noted that initially, briefly, the Court was minded to interpret the EU Treaty through ‘a more conserving approach’ (which would certainly fall within VCLT). ¹²³ Instead, the teleological approach became entrenched remarkably quickly and EU law developed in a significantly different fashion as a result.

The reality is that differences between the methods applied can have a decisive impact on the outcomes of cases, at least in EU courts. ¹²⁴ With VCLT methods suggested to be more conservative in practice than the EU teleological method, a note of caution is warranted concerning ab initio preferring of VCLT on the basis it better serves legal certainty.

First, we have identified that justice is also a meritorious pursuit. Accordingly, often those who have criticised the Court’s teleological approach for its impact on legal certainty have mixed this with their conception of justice. For example, Rasmussen states ‘[p]ro-integration judicial activism should be rejected when union-promotion takes place at the expense of the EC’s other fundamental values, including legal certainty’. ¹²⁵ However, concerning fundamental rights protection (an example of meta-teleological interpretation) Rasmussen argues the Court should have taken action sooner. ¹²⁶ Others (including the Court) may draw different lines for legal certainty based on their view of justice.

Second, it is notable that many scholars also appeal to legal certainty to support the Court’s teleological approach. ¹²⁷ After careful study, Beck concludes that the Court’s decisions have ‘a reasonable degree of reckonability’ and ‘may well exceed the predictability of many decisions in many national constitutional and final appeal courts.’ ¹²⁸ Elaborating further, Maduro highlights key benefits of the teleological approach ¹²⁹ :

It forces the Court to highlight its normative understanding of the EU legal order and it creates a yardstick to better assess the coherence and consistency of its case law. It also creates an opportunity for a broader debate on the nature of the EU legal order and its underlying values, while more formal forms of legal reasoning would hide discretion and preclude debate.

Whilst references in this passage are made to EU law, bilateral and multilateral treaties and case law analysing the impacts of treaties on third parties would benefit similarly from inherent aspects of this interpretative method.

With plausible arguments attributable to the justice and legal certainty merits within both VCLT and teleological methods it is not proposed (nor possible) to favour one method in abstract. Each may have its appropriate place and its own appropriate execution. What follows is an analysis of the contexts of these: When and how does the Court select and apply each method? And how does this achieve or detract from legal certainty and/or justice?

The following sections will analyse the case law across the strands of bilateral treaties, multilateral treaties and treaties impacting third parties. It will consider each of the relevant strands within a loose division of ‘thin’ justice, ‘thick’ (or substantive) justice and legal certainty. Each case will inevitably have consequences for ‘thin’ justice, ‘thick’ justice and legal certainty, but this loose division will enable us to focus more closely on these key aspects within the case law.

4. ‘Thin Justice’ emphasis: selecting and applying VCLT or teleological interpretation

For treaties impacting third parties the Court has consistently found that only VCLT and customary international principles are relevant for interpretation. In contrast, for bilateral and multilateral treaties either VCLT or teleology may be applied, raising questions for determining which will be applied when? As such, the ‘thin’ justice analysis here excludes treaties impacting on third parties and focuses on bilateral and multilateral treaties.

Treaties hold the prospect, recognised under international law itself, that dynamic interpretative approaches can be established through subsequent practice ¹³⁰ or that this may have been intended from the outset based on context and purpose. ¹³¹ This possibility is clearly most relevant to bilateral treaties, which form the vast bulk of EU international agreements. ¹³²

Given the prominence of VCLT, where the contracting parties have not indicated in the text the prospect of innovation or teleology it is unclear as to when VCLT should be moved away from (especially in multilateral agreements and more limited bilateral agreements eg free trade and co-operation compared to Association ¹³³ ). Ultimately, we will see that while both teleological interpretation and VCLT may be applied to bilateral treaties, teleology appears to be expanding. However, the logic as to agreements which fall within or beyond its reach is unclear. It will also be seen that while teleology may not be expected to be relevant in the multilateral context, the EEA agreement provided a foothold for teleology in this sphere.

Flexibility in selecting interpretative approaches prioritises the Court’s capacity to deliver substantive justice in each individual case over legal certainty, but it may not currently be executed in achievement of ‘thin’ justice. The lack of consistency in differentiating cases on the basis of ‘thin’ justice also inevitably raises questions as to the overarching substantive justice concept(s) guiding the Court, in addition to further impacting legal certainty. This can now be explored in the case law.

A. Bilateral treaties

The Polydor principle

In Polydor ¹³⁴ the Court was confronted with a free trade agreement ¹³⁵ containing provisions which ‘in several respects are similar to those of the EEC Treaty on the abolition of restrictions on intra-Community trade.’ ¹³⁶ The case covered a topic familiar to EU law: free movement of goods and their restriction. Nonetheless the Court found that ‘such similarity of terms is not a sufficient reason for transposing to the provisions of the agreement the above-mentioned [internal] case-law'. ¹³⁷ The result was that ‘restrictions on trade in goods may be considered to be justified on the ground of the protection of industrial and commercial property [in this case musical copyright] in a situation in which their justification would not be possible within the Community.’ ¹³⁸ This was found to be appropriate as ‘the Community’s objectives and activities …. seeks to unite national markets into a single market having the characteristics of a domestic market.’ ¹³⁹ In short, because the Portuguese agreement reflected a lesser level of integration than the EU Treaty, the Court was less intrusive in removing barriers to free movement of goods than it would have been internally.

However, in Commission v Italy ¹⁴⁰ the Court was called upon to interpret an agreement for which the Commission had previously noted the ‘general provisions and the object are, in principle, the same’. ¹⁴¹ As with the Portuguese agreement in Polydor, the agreement here with Norway ¹⁴² was also a ‘‘mere’ free trade agreement’. ¹⁴³ Commission v Italy, then, seemed not to be a promising case for applying the teleological approach. However, the Court held – without making any reference to Polydor – that ¹⁴⁴ :

[T]he Agreement contains with respect to trade between the Contracting Parties rules identical to those of Articles 30 and 36 of the Treaty and there are no grounds in this case for interpreting those rules differently from those Articles of the Treaty.

The consequence was that the barrier to free movement of goods could not be justified, resulting in a significantly different outcome to Polydor due to an approach that was distinct. One adhered to teleology, the other was in closer accordance with VCLT methods (although not cited).

Analysis of the text of the agreements is limited in each of these two cases, but this could not account for the difference in treatment given their similarity. Concerning the parties to the agreements, when Polydor and Commission v Italy were decided in 1982 and 1993 respectively, Portugal had formally requested to join the EU in 1977, had approval from the Council and EU Parliament by 1982 and would do so in 1986, conversely Norway was just a year away from confirming (for a second time) through referendum in 1994 that it would not join. Objectively it seems that Portugal was in at least as close a relationship with the EU as Norway.

The difference may be accounted for in the distinction between the subject matters interacting with free movement of goods; exhaustion of rights and health protection. But the determinative difference between the two is not clear. ¹⁴⁵ AG Rozés’s suggestion in Polydor that exhaustion of rights comprises or anticipates not merely free movement of goods but integration of services, and possibly establishment, is interesting. ¹⁴⁶ These criteria, though, would also pose challenging questions as to which free movement of goods barriers would fall within that remit. The Court has been less forthcoming, and outcomes have varied. Matters of taxation may not require such integration yet have not always been afforded the extension of teleological interpretation. ¹⁴⁷ A further contributing (if unhelpful and unexpressed) factor differentiating our cases may be that in one the party invoking the international treaty is the Commission, making robust enforcement more likely. ¹⁴⁸

The possibility that similar or identical provisions in international agreements may require the same interpretation as those given to EU Treaty provisions has come to be known as the Polydor principle. But it has been suggested that Polydor ‘provides no firm guidance on the conditions that need to be met for such similarly worded provisions to be interpreted in a similar manner’. ¹⁴⁹ Indeed, AG Rozés’ approach suggests the Polydor constellation can even call for a duality of interpretative techniques within each provision depending on specific factual matrices. This is a very highly calibrated – perhaps over-calibrated – effort to distinguish cases. Even if this is the aim, recalling that justice does not act as a cap on all forms of legal certainty, the Court could have proffered a fuller rationale in Polydor and at least revisited/referenced its scope in Commission v Italy.

The Court’s discretion in applying the Polydor principle

There has been some further elaboration from the Court concerning the Polydor principle. The Court states the extension of internal interpretation ‘depends, inter alia, on the aim pursued by each provision [of an international agreement] in its own particular context’ and that ‘comparison between the objectives and context of the agreement and those of the Treaty is of considerable importance in that regard’. ¹⁵⁰ However, the extent to which the Court’s application of these criteria has provided a more solid basis on which to differentiate cases (‘thin’ justice), emerges as highly questionable. Key aspects concerning agreement types, the relationship of the EU with relevant partners and even the texts of agreements themselves can be accentuated or ignored by the Court, as we will now explore.

Given the importance of parties’ intentions and the text of the agreement in international interpretation, one might expect that a clear direction concerning interpretation within an agreement would be the most relevant aspect in determining how to proceed. ¹⁵¹ The Court has acknowledged that it will conform with this principle of international law, meaning parties to a treaty are ‘free to agree’ the domestic effects of concluded agreements. ¹⁵² Yet teleological interpretation is not always expressly provided for but is at times applied. This raises questions as to how it can emerge from the ‘objectives and context’ test attributed to Polydor, and dislodge the default expectation that VCLT be applied. Conversely, where equivalence with internal interpretation has been provided for in the Agreement, the Court has not always followed this instruction.

Concerning the Association Agreement with Turkey ¹⁵³ the Court draws attention to, and follows the instruction, that under the Association Agreement ‘the Contracting Parties agree to be guided by’ relevant EU Treaty provisions. ¹⁵⁴ However, such references to internal principles have proved to be non-dispositive concerning the Swiss agreements. In both Grimme ¹⁵⁵ and Fokus Invest ¹⁵⁶ Article 16(2) of the relevant Swiss agreement provided ¹⁵⁷ :

Insofar as the application of this Agreement involves concepts of Community law, account shall be taken of the relevant case-law of the Court of Justice of the European Communities prior to the date of its signature.

Both cases turned on whether legal persons were granted rights of establishment under the agreement in addition to natural persons, as is the case internally. ¹⁵⁸ The Court declined to extend internal rules in spite of Article 16(2), however, given that specific provisions were present, ¹⁵⁹ arguably broader internal concepts were of less relevance.

More controversially though, in Hengartner and Gasser, in spite of referring similarly to Article 16(2), ¹⁶⁰ the Court cited the Vienna Convention and maintained ¹⁶¹ :

[T]he interpretation given to the provisions of European Union law concerning the internal market cannot be automatically applied by analogy to the interpretation of the Agreement, unless there are express provisions to that effect laid down by the Agreement itself…

Accordingly, even though the agreement referred to ‘establishment’ (thereby prompting the referring court to draw equivalences with the EU internal concept) the Court refused in principle to accept any equivalence. ¹⁶² The 2018 Picart case provides confirmation of the Court’s exclusion of teleology (but without making specific reference to VCLT). In N, an internal case, it had accepted that establishment could be recognised where significant shareholding in a company was present but work (decision-making) was undertaken from another Member State as the definition of establishment ‘is a very broad one’. ¹⁶³ In Picart the Court refused to apply this and, as in Hengartner, the need for an express provision on interpretation was emphasised. ¹⁶⁴ Anticipating an express provision in each instance – a requirement which applies beyond establishment alone ¹⁶⁵ – precludes Article 16(2) of having any meaningful effect.

The prospect of broad certainty (if not necessarily justice) in Swiss agreements consistently being denied teleological interpretation where others are not waivers slightly with the Court accepting the need for such interpretation in coordination of social security systems. ¹⁶⁶ This relates not directly to the text of agreements with Switzerland but to interpretation of EU secondary legislation; here Switzerland had been brought within the scope of the definition ‘Member State’ for those purposes specifically through Article 8 of the EC-Switzerland Agreement. It is also worth noting that this exception to the trend in case law rendered unsuccessful a UK challenge to annul EU legislation and seek costs. ¹⁶⁷

A key difference in a more restrictive approach to the Swiss agreements compared to others ¹⁶⁸ may be that in most Swiss cases the Court has highlighted – somewhat sourly – the fact that Switzerland opted not to join the internal market project. ¹⁶⁹ The relevance of declining to ratify a separate treaty when deciding upon another treaty’s interpretation is questionable. ¹⁷⁰ It is also inconsistent with previous case law (at the time of deciding Commission v Italy, Norway had already previously opted similarly and that did not preclude teleological interpretation being applied). It is also unclear why in Hengartner and Gasser the Court referred only to the Vienna Convention concerning Article 31(1) (wording and purpose) but did not cite Article 31(4), providing that a special meaning shall be given if the parties intended. Arguably this was not just a refusal to extend teleological methods to the agreement, but simultaneously a denial of VCLT methods too.

Teleological interpretation has been applied particularly to Association Agreements. A common feature of older versions of such agreements is that ‘all the associated partners under these agreements aspired or aspire to become members of the EU’. ¹⁷¹ When put in those terms it perhaps seems reasonable to fast track the (presumably) desired outcome for both parties and to render the agreement, with its similar provisions, as close to EU law in action as possible so as to ease transition. ¹⁷² However, if the parties intended this a counter-argument is that it can be expressly provided for in the agreement.

Interestingly, certain provisions of Association Agreements have been viewed by the Court as not allowing for teleological interpretation, particularly concerning rights of residency. Here the Court has on occasion stated that the Association Agreements are ‘designed simply to create an appropriate framework for [eg] the Republic of Poland’s gradual integration into the Community, with a view to its possible accession, whereas the purpose of the Treaty is to create an internal market’. ¹⁷³ This, though, has marked the exception to a general trend of extending teleology.

In Pokrzeptowicz-Meyer, ¹⁷⁴ concerning the same Association Agreement with Poland, ¹⁷⁵ the Court found no problem in extending (now) Article 45(2) TFEU interpretations concerning non-discrimination against workers to Article 37(1) of the Agreement. Unlike the Turkey Association Agreement, or even the Swiss Agreement, there was no clear indication of being ‘guided’ by or ‘taking account of’ EU Treaty provisions and case law in the Polish Agreement except in the field of competition law. ¹⁷⁶ The relevant provision concerning workers in the case, Article 37(1), instead used the phrasing ‘[s]ubject to the conditions and modalities applicable in each Member State’, which German authorities claimed introduced conditionality. ¹⁷⁷

The Court held that the provision could not be interpreted so as to allow non-discrimination to be subject to ‘conditions or discretionary limitations’ as these would ‘render the provision meaningless’. ¹⁷⁸ As a result the rigorous internal approach to indirect discrimination was applied to the agreement. ¹⁷⁹ German rules allowing for fixed-term contracts to be used for foreign-language assistants were compared to ‘other teaching staff performing special duties’, where an objective reason for such practice was required. ¹⁸⁰ Whilst foreign-language posts were also occupied by German nationals, the Court transposed its reasoning on the same provision of German law in the internal case of Spotti, ¹⁸¹ where it was claimed that ‘a great majority’ ¹⁸² were foreign nationals. Accordingly, there was evidence of indirect discrimination in this case, ¹⁸³ which the Court accepted as being sufficient to breach the requirement of no discrimination on the basis of nationality.

The extension of internal case law was less than predictable in Pokrzeptowicz-Meyer. The previous tone adopted in Gloszczuk just four months earlier was jettisoned as the Court appeared to recast the aims of the Association Agreement from ‘gradual’ to ‘progressive’ integration, ¹⁸⁴ with the qualification ‘possible’ omitted concerning accession. The term ‘subject to conditions and modalities’ could also have been more convincingly addressed by the Court. It differs from Article 45 TFEU and undermines the Court’s claim that Article 37(1) would be meaningless without equally broad interpretation.

This bolder approach has not been limited to Association Agreements. In Kziber the Court noted that the agreement with Morocco ¹⁸⁵ ‘confines’ itself to cooperation and did not anticipate future accession. ¹⁸⁶ It was thus a ‘“mere” co-operation agreement’, then ‘the weakest type of bilateral agreement between the EU and a third country.’ ¹⁸⁷ The Court, though, found no problem in the concept of a worker being extended to those who ‘have left the labour market after reaching the age required for receipt of an old-age pension’. ¹⁸⁸ Apap notes that this reading was more extensive than case law concerning the definition of workers under Decision 1/80 of the EU-Turkey Association Agreement to that date, ¹⁸⁹ instead fitting with internal case law. ¹⁹⁰ Thus, provisions in more ‘confined’ agreements may occasionally be interpreted as extensively as they would be internally without waiting for a trickle-down effect from more integrationist agreements. ¹⁹¹ This is especially problematic concerning ‘thin’ justice as it is hard to understand when an agreement may spontaneously be elevated beyond that which has gone before.

B. Multilateral treaties

Multilateral treaties are not obvious candidates for extension of the EU’s teleological interpretative method. First, the text of multilateral treaties is less likely to be similar to that of the EU Treaty, which was the starting point for the Polydor principle. Second, teleology in bilateral agreements has experienced a strong trickle-down effect from Association Agreements. Third, with a greater number of parties the prospect of subsequent practice establishing agreement concerning interpretation lessens. ¹⁹² This entrenches VCLT and makes attempts to move away from it often appear less than bona fide. ¹⁹³ However, extension of internal teleological interpretation to multilateral treaties has at times featured, starting with the European Economic Area Agreement (EEA). ¹⁹⁴

In Opinion 1/91 (initially refusing to approve the EEA agreement) the Court stated that due to ‘the divergences which exist between the aims and context of the agreement’ ¹⁹⁵ it would be necessary to apply Article 31 VCLT to it. ¹⁹⁶ The proposal for EU judges to sit in the EFTA Court was rejected as it would be ‘impossible’ for judges to switch between different interpretative methods with ‘completely open minds’ ¹⁹⁷ , a point which may raise a few eyebrows as the Court shuttles between teleology and VCLT across its international case law.

The Court subsequently approved the agreement in Opinion 1/92. ¹⁹⁸ The new agreement continued to seek ‘as much homogeneity as possible’ but stopped short of creating a combined court. ¹⁹⁹ In Opel Austria the General Court dutifully interpreted Article 10 of the EEA agreement in parallel with the EU Treaty (the Article concerns customs duties and charges having equivalent effect, and is very similar in wording to (now) Article 30 TFEU). ²⁰⁰ This followed the requirement of Article 6 EEA to interpret EEA provisions ‘in conformity with the relevant rulings of the Court of Justice’. The Court has continued this practice. ²⁰¹ Albeit, perhaps unsurprisingly given the Court’s observation in Opinion 1/91 that ‘it will be difficult to distinguish the new case-law from the old and hence the past from the future’, ²⁰² it appears subsequently to have gone beyond the Article 6 EEA requirement to only apply judgments ‘prior to the date of signature’. ²⁰³

This impacts upon EFTA members who ‘did not wish to commit themselves to future, unforeseeable developments in the case law’ ²⁰⁴ and, of course, on EU institutions and particularly the Member States (against whom most cases have arisen). However, ‘the EFTA Court has in practice never made a distinction between pre- and post-signature case law of the ECJ’ ²⁰⁵ and, given the strong emphasis on homogeneity in the EEA, ²⁰⁶ some argue ‘there would have to be strong reasons for not extending EU law interpretations.’ ²⁰⁷

Through case law concerning the EEA the principle that teleology could be applied to multilateral treaties was established by the Court, albeit one might not have thought it would be of much relevance beyond this very specific context.

C. Appraisal

Writing in 2008 Kaddous stated that ‘there are still differences…and it remains difficult to determine to what extent the nature of agreements will be taken into consideration [for the purposes of interpretation]’. ²⁰⁸ Over a decade on Ghazaryan writes ²⁰⁹ :

[H]aving an element of EU acquis transposition itself does not guarantee similar interpretation. Neither does a promise of association always secure a homogenous interpretation. It is, therefore, difficult to deduce a clear and consistent pattern in the Court’s case law.

These are conclusions which support a thesis in which little progress has been made to articulate criteria for ‘thin’ justice in this area.

Of course, pursuit of substantive justice may justify shortcomings in legal certainty. Accordingly, Koutrakos has previously acknowledged this ‘uncertainty’ but that some flexibility may be ‘positive’ in ensuring that the ‘judiciary be free to take any developments into account when interpreting international rules.’ ²¹⁰ This is undoubtedly so, and we have seen the Court find it necessary to expand the option of teleology beyond bilateral agreements to the multilateral EEA Agreement.

However, we have also noted that flexibility, or justice, does not operate as an absolute cap on legal certainty in many instances. While the Court has provided some rationale and continued dialogue for teleology concerning the EEA Agreement, it could certainly be more forthcoming in its reasoning concerning selection and execution of interpretative methods for bilateral agreements. In absence of this, meta-narratives have been identified by scholars – stronger enforcement of international treaty law against Member States, ²¹¹ a reluctance to advance teleological interpretation to rights related to residence ²¹² – but the Court does not acknowledge them, let alone integrate them as clear Polydor criteria.

Without the Court indicating factors that account for dissimilarities between cases speculation has been necessary and, having narrowed our field of analysis, it feels like we are well placed to at least question whether formal or ‘thin’ justice is being satisfactorily served in aspects of this line of case law. Clearly this also impacts on legal certainty as without knowing ‘what differences are relevant’ ²¹³ for the purposes of ‘thin’ justice, the aim that ‘individuals may ascertain unequivocally what their rights and obligations’ cannot be met. ²¹⁴ Persistent shortcomings in ‘thin’ justice also indirectly pose questions concerning ‘thick’ or substantive justice. Relevant differences will vary depending on the overarching purpose or aim; if this aim is consistent, one might expect that law would ‘work itself pure’ over time. ²¹⁵ However, in subsequent section we will consider further case law, with an emphasis on substantive justice, and suggest that tentative concerns for the lack of an overarching vision may be well founded.

5. Substantive justice emphasis: teleology’s over-extension and covert deployment

Each of the cases analysed in the context of ‘thin’ justice above will, of course, have implications for legal certainty. Each also had a substantive outcome. This section provides opportunity to consider further case law with an emphasis on substantive justice (noting once more that these cases will also impact certainty and ‘thin’ justice).

Challenges in identifying a coherent approach to ‘thin’ justice imply that there may be lacking a consistent overarching vision of substantive justice. In our example concerning twin children’s bedtime we considered activity in each day to be relevant but the colour of clothing worn to be irrelevant. This ‘thin’ justice judgement reflected our deeper (substantive) purpose in ensuring the children would sleep an appropriate amount. Considering key case law here alongside the substantive issues at stake will provide some context for outcomes. Of course, substantive justice will remain debatable by its very nature. However, it is submitted that the findings will raise some concerns.

We will see that in bilateral case law the Court is willing to extend teleology to even the least integrationist of bilateral agreements, basing reasoning on the most unconvincing of ‘thin’ justice arguments. In the multilateral context the Court appears unwilling to extend teleology formally beyond EEA (through arguments based on ‘thin’ justice) but instead appears to deploy it covertly, based, we may speculate, on a vision of substantive justice (notably for consumer protection). Finally, the Court’s case law concerning treaties impacting on third parties will be addressed; there the Court does not entertain ‘thin’ justice distinctions in determining whether to apply teleological or VCLT interpretation (hence this topic’s omission from the previous section). Instead, it purports to apply only international interpretative methods, but, in reality, the approach will be argued to be vacuous as concerns substantive justice.

A. Bilateral treaties

Scholars have noted that (Switzerland aside ²¹⁶ ) a broad trend of extending internal interpretations to international agreements has continued ²¹⁷ and Simutenkov ²¹⁸ arguably marks its peak. The case concerned a Partnership and Cooperation Agreement (PCA) with Russia. ²¹⁹ This type of agreement arguably creates even more limited relations between the parties than the older co-operation agreement of the type concluded with Morocco (discussed above concerning the Kziber case ²²⁰ ), being described as an ‘entry-level agreement’, ²²¹ ‘on the whole rather meagre in terms of vision and concrete content.’ ²²²

With Morocco, the agreement was of unlimited duration and provided for a wide area of cooperation whereas PCAs, such as that concluded with Russia, anticipate running for a ten year period, after which renegotiation is needed. ²²³ Concerning the Russian agreement, Koutrakos draws particular attention to the emphasis on ‘mutual advantage, mutual responsibility and mutual support’ in substantive areas. ²²⁴ In the agreement with Morocco there was instead asymmetry, ²²⁵ which arguably justifies a more purposive (teleological) approach to its provisions.

However, in Simutenkov, concerning restrictions on professional football teams fielding non-EU players, these factors did not prevent ‘the active transposition of notions of EU substantive and constitutional law’. ²²⁶ The Court was facilitated in its reasoning due to determining the effects of a very similar provision in the Association Agreement with Slovakia. ²²⁷ In Kolpak a handball player challenged the decision to grant him a licence which referred to his non-EU status as only two such players could be fielded by each team. In answer the Court drew close analogies with Bosman applying (now) Article 45(2) TFEU concerning non-discrimination of workers on the basis of nationality. This meant Simutenkov saw the application of internal EU law by extending Kolpak, which was itself an extension of the Bosman case, ²²⁸ with the justification and capacity to anticipate application of the teleological approach diluted with each extension. ²²⁹

The outcome of Bosman is well known; sport can amount to an economic activity and be subject to EU law. ²³⁰ As per previous case law Article 45 TFEU ‘not only applies to the action of public authorities but extends also to rules of any other nature aimed at regulating gainful employment in a collective manner’ ²³¹ and the restriction on foreign nationals from other Member States, if allowed, would deprive the Treaty of its ‘practical effect and the fundamental right of free access to employment which the Treaty confers individually on each worker’. ²³²

In Kolpak the Court established for the first time that an international agreement could have horizontal direct effect, allowing even private entities to be bound by it. ²³³ In spite of such a significant step, the Court found no ‘objective justification for the difference in treatment between, on the one hand, professional players who are nationals of a Member State or of an EEA Member State’ ²³⁴ due to ‘the aims and context of the Association Agreement’. ²³⁵ The Court expressly laid out the objectives of the Agreement as ²³⁶ :

[E]tablishing an association to promote the expansion of trade and harmonious economic relations between the contracting parties so as to foster dynamic economic development and prosperity…in order to facilitate those countries’ accession to the Communities.

The result was that the reasoning from Bosman could be fully ‘transposed’ onto Article 38(1) of the Association Agreement. ²³⁷

In contrast, to dynamism and accession, in Simutenkov it will be recalled that the Court was confronted with a markedly more cautious agreement. In spite of this it noted that Article 23 of the PCA, requiring equal treatment of Russian nationals with Member State nationals as concerned inter alia working conditions, was ‘very similar’ to Article 38(1) of the Slovakian agreement in Kolpak. ²³⁸ This preference for redeploying matters decided in a previous case, in a very different context, ²³⁹ arguably adds ‘convenience’ and ‘coincidence’ to the Polydor pantheon. It also marks a departure from statements in El-Yassani, where the Court was robust in refusing to draw equivalences between Association Agreements and free trade agreements (although El-Yassani itself differs from Kziber, where teleological interpretation was applied to the same agreement). ²⁴⁰

Unable to cite broad and ambitious aims of the agreement with Russia, the reasoning in Simutenkov was even less clear than Kolpak ²⁴¹ :

Admittedly, unlike the Communities-Slovakia Association Agreement, the Communities-Russia Partnership Agreement is not intended to establish an association with a view to the gradual integration of that non-member country…

However, it does not in any way follow from the context or purpose of that Partnership Agreement that it intended to give to the prohibition of ‘discrimination based on nationality, as regards working conditions […]’ any meaning other than that which follows from the ordinary sense of those words.

Appealing to ‘the ordinary sense’ of the words is itself questionable given that Mendez describes the Bosman ruling as ‘seminal’ and ‘the quintessential of judgments in need of a fundamental freedoms underpinning’. ²⁴² The emphasis in the agreement with Russia on mutuality also suggests that an overlay of some caution was warranted.

In this regard, Maresceau notes that ‘attempts by Russia during the course of the negotiations to reach a bilateral framework that was somehow comparable with [Association Agreements] met with strong opposition on the EC side.’ ²⁴³ Accordingly, he emphasises that PCAs ‘are a far cry from being an alternative to the [Association Agreements]’ ²⁴⁴ and attempts by the Court to treat them as equivalent are noted to be ‘remarkable’. ²⁴⁵ Yet, increasingly, their provisions are not interpreted differently from Association Agreements and, therefore, the EU Treaty itself.

Suffice to say the reasoning in Simutenkov has not aged well. Even before Russia’s full-scale invasion of Ukraine in 2022, the Court had already been called upon to interpret the EU’s relationship with Russia once more, this time to confirm the discretion of the Council in imposing sanctions and their permissibility under the PCA. ²⁴⁶ The EU’s political relationship with Russia has variously been cautious, co-operative, competitive or mired in crisis. ²⁴⁷ Simutenkov marks the danger of the Court joining in the politics, departing from more anticipatable interpretative criteria and impacting legal certainty to pursue a vision of substantive justice which does not endure.

B. Multilateral treaties

Whilst the Court had established the principle that teleological reasoning could be appropriate in the multilateral context concerning the EEA agreement, this seemed to be the exception. Indeed, the case of ELFAA, ²⁴⁸ concerning a challenge to an EU Regulation on the basis that it breached the 1999 Montreal Convention concerning air carrier liability, proved markedly controversial in extending teleological interpretation to a further multilateral treaty, especially as this was done covertly.

The Montreal Convention provides for the circumstances in which passengers are entitled to compensation and limits the amount. The EU is party to the Montreal Convention and the Court found the relevant Articles of the Convention to have direct effect. ²⁴⁹ It then referred to Article 31 of the VCLT and directed itself that an international treaty ‘must be interpreted … in good faith in accordance with the ordinary meaning to be given to its terms in their context and in the light of its object and purpose.’ ²⁵⁰

The Court acknowledged that the Convention limited liability of carriers to ‘4 150 [Special Drawing Rights (SDRs)] for each passenger’. ²⁵¹ However, it claimed ‘[t]he system prescribed in Article 6 [of the Regulation] simply operates at an earlier stage than the system which results from the Montreal Convention.’ ²⁵² The challenge was unfounded; the EU legislature was free to impose liability on carriers.

The Regulation covered ²⁵³ :

[S]tandardised and immediate assistance or care for everybody concerned, through the provision, for example, of refreshments, meals and accommodation and of the opportunity to make telephone calls.

The Convention, the Court argued, covered individual damage sustained by passengers. The judgment implicitly acknowledges the potential for some overlap with the Convention ²⁵⁴ and clearly some expenses now dealt with under the Regulation would previously have formed damages under the Convention (the prudent traveller would have claimed for ‘refreshments, meals and accommodation’ etc.). The 4 150 SDRs cap is therefore essentially raised, in breach of the Convention’s exclusivity provision (allowing for passengers to seek damages only through the Convention). ²⁵⁵

Due to its technical, specific and pan-global nature, the Montreal Convention has been observed to be among those rules where international uniformity – as oppose to localised development – would be expected. ²⁵⁶ Truxal draws attention to the fact that in contrast to the CJEU’s ‘finesse’ of the Convention, US courts have found the Convention to be unequivocal in setting a common standard and have diligently followed it, ²⁵⁷ as has the House of Lords (now UK Supreme Court). ²⁵⁸ With the Court’s approach marked as an outlier – ‘the predictability that adherence to the treaty has achieved worldwide’ was noted in US case law ²⁵⁹ – and the merits for extending teleology in this case appearing limited, it is not surprising that there was also broad criticism of the ELFAA case from scholarship. ²⁶⁰

ELFAA appears to be an effort from the Court to pursue the substantive aim of consumer protection in the field of aviation. This concern had been evident in the internal case of Sturgeon ²⁶¹ where, in spite of Regulation 261/2004 ²⁶² appearing to treat events as distinct, the Court equalised the entitlement of airline passengers to compensation in the case of delay of more than three hours to that of cancellation. Of course, flexibility is provided by the teleological approach to interpretation and pursuit of consumer protection is provided for in the EU Treaty itself. ²⁶³ However, aside from questions concerning how this approach may translate to the international context, in ELFAA the Court purported to apply VCLT, not teleology.

It is also notable that in a line of case law beginning with Walz, concerning damage to baggage, the Court sought to lay emphasis on the fact that the Montreal Convention seeks ‘an ‘equitable balance of interests’ to be maintained, in particular as regards the interests of air carriers and of passengers.’ ²⁶⁴ This observation was absent from the Court in ELFAA, where Article 22(1) of the Convention liability cap for delay was essentially disregarded. In Walz and subsequent case law, however, the Article 22(2) cap concerning baggage was respected and surprisingly endorsed ²⁶⁵ :

Indeed, a limitation of the compensation so designed enables passengers to be compensated easily and swiftly, yet without imposing a very heavy burden of damages on air carriers, which would be difficult to determine and to calculate, and would be liable to undermine, and even paralyse, the economic activity of those carriers.

Accordingly, the Court refused to treat material and non-material damage as distinct in the same way it had standardised and individual damages in ELFAA.

One wonders whether the substantive justice aim of protecting (many) passengers from delay compared to (far fewer) from damaged luggage may have caused greater boldness from the Court in pursuing the former. If so, it is clear that this is not articulated and that the current ‘thin’ justice construct struggles to take this ‘thick’ justice consideration into account. Moreover, the tone – and that of subsequent judgments ²⁶⁶ – endorses the Convention to such a great extent that we may question whether hypothetically the Court would decide ELFAA differently today. This is not a change caused by external supervening events (unlikeness) but is instead a result of the Court’s conception of justice changing in this area. Given the uncertainty of the interpretative method applied in ELFAA (clearly not VCLT), should the substantive justice conception it pursued not stand, it becomes arguable that ELFAA is the exemplar of a case that is uncertain and unjust.

C. Treaties impacting third parties

It has been suggested that bilateral treaties had generally been given the widest possible (teleological) interpretation. This was subject primarily to the exceptions of the Swiss agreement and provisions relating to residence. In addition to some fresh caution in the multilateral context, there is a further limitation to providing interpretations based on the EU teleological method. This concerns treaties which potentially impact on the rights of third States or entities. In this line of case law, the Court eschews teleology and expressly links its reasoning to VCLT and relevant customary international law principles. As such this topic was not included in the previous section on ‘thin’ justice, as the Court does not formally entertain selecting between international and EU interpretative methods in this area.

This is a topic which has been recurring, with case law before the General Court and/or Court in 2010, ²⁶⁷ 2015, ²⁶⁸ 2016, ²⁶⁹ 2018, ²⁷⁰ and 2021 (featuring two judgements on a single day, ²⁷¹ both of which are currently under appeal ²⁷² ). Whilst many cases have focused on Western Sahara, it has also been suggested that, given the EU’s treaty making practices and contested territories in its neighbourhood, this area will be of increasing relevance moving forward. ²⁷³ The extent to which the Court appropriately deploys international principles in this area is increasingly questioned, meaning it is likely this area of law will continue to attract (negative) attention.

Article 34 VCLT provides that ‘[a] treaty does not create either obligations or rights for a third State without its consent’. The Court has cited this Article alongside the principle of self-determination (an erga omnes responsibility in international law) ²⁷⁴ in order to apparently mark limits for the consequences of treaties entered into by the EU. Brita provided indication that the Court considers Article 34 to be appropriate in considering the ‘context’ of a treaty for the purposes of Article 31 interpretation. ²⁷⁵ On this basis the Court restricted the scope of application of the EC-Israel agreement, also noting that only in this way could an existing EC-PLO Protocol continue to have meaningful effect (essentially focusing on the roles allocated to customs authorities rather than more contentious issues). ²⁷⁶

In its caution the Court’s approach generally has differed from that of the General Court; in Front Polisario the General Court considered that the Council decision implementing the agreement needed to be annulled insofar as it covered Western Sahara, even citing an obligation for the Council to ‘examine all the elements of the case before the adoption of the contested decision’. ²⁷⁷ However, the Court was not convinced such an obligation existed. Nor was it swayed by evidence that there was an understanding and de facto practice between the EU and Morocco in implementing an Association Agreement so as to cover the territory of Western Sahara. ²⁷⁸ Indeed, further evidence that the Council and Commission intended to cover the contested territory was ignored by the Court. ²⁷⁹ Accordingly, the Court was far more conservative in its approach and instead interpreted a presumptive compliance with international law, rather than considering actual intent or conduct, a trend which has continued. For these reasons it has been suggested that the Court’s approach in this area does not reflect a balanced approach towards the elements of interpretation presented in Article 31 VCLT. ²⁸⁰

These criticisms appear pertinent once more in Western Sahara as, rather than focusing on the text of the agreement or parties’ intentions, the Court stated: ²⁸¹

If the territory of Western Sahara were to be included within the scope of the Association Agreement, that would be contrary to certain rules of general international law that are applicable in relations between the European Union and Kingdom of Morocco, namely the principle of self-determination, stated in Article 1 of the Charter of the United Nations, and the principle of the relative effect of treaties, of which Article 34 of the Vienna Convention is a specific expression.

Due to its obligations under international law ‘the European Union could not properly support any intention of the Kingdom of Morocco to include, by such means, the waters in question within the scope of that agreement.’ ²⁸² The outcome in Western Sahara was, to an extent, unsurprising given that even where there was evidence of ‘de facto’ application of an agreement to a contested region in Front Polisario the Court circularly found ‘the purported intention of the European Union, reflected in subsequent practice’ ²⁸³ could not be to breach international law as ‘such implementation would necessarily be incompatible with the principle that Treaty obligations must be performed in good faith’. ²⁸⁴

In terms of substantive justice, the Court’s approach can be argued to be ambivalent or vacuous. It appears not to entertain real world impacts and instead focuses on sophistry. We shall see further below that strong criticism – even beyond academic texts – is increasing pressure on the Court to change path. ²⁸⁵ This has mainly arisen concerning the lack of adherence to international interpretative standards, suggesting that there is little merit in the cases from the external perspective. Internally, the Court is also coming dangerously close to undermining its conception of the Rule of Law, a principle on which the EU is founded. ²⁸⁶ In Les Verts the Court described the fact that neither the Member States nor the EU institutions could avoid their actions being reviewed on the basis of the Treaty as a fundament of the ‘Community based on the Rule of Law’. ²⁸⁷

D. Appraisal

We have previously identified that ‘thin’ justice has not emerged clearly in the case law and in the next section we will go further in identifying challenges concerning legal certainty. With those two elements falling short, significant pressure is placed on the Court to always achieve a substantive outcome which is satisfactory to participants. This is clearly a challenging task as views on substantive justice often differ greatly and change over time.

There is not sufficient space to capture in detail the evolutions in the EU's – and the Court’s – substantive justice sentiments. However, we can reflect on the fact that whilst the EU was initially a trade-focused entity, today its pluralistic pursuits include inter alia trade, human rights, environmental protection and consumer protection: multiple competing interests which must be resolved in each policy and case. ²⁸⁸ Internally proportionality has been a helpful tool in the Court’s incremental management of these aims. ²⁸⁹

In case law, applying international law proportionality is largely absent. ²⁹⁰ Instead, from an early stage, the Court committed to automatically incorporating international treaties as an ‘integral part’ of EU law and to affording them a rank above secondary legislation. It similarly asserted that the EU ‘must respect customary international law in the exercise of its powers’. ²⁹¹ Had the Court not adhered to international law from the outset, its demands that Member States respect its ‘new legal order of international law’ would likely have rung hollow. Today, though, the Court is significantly less reliant on international law to legitimise itself. ²⁹² It has also developed an internal body of case law which reflects diverse norms of EU law. Inevitably these will not always sit well with the EU’s international obligations.

As such, prominent cases in which the Court has been criticised for its approach to international law have concerned familiar internal norms. It is not surprising that the first tensions between EU and international law concerned trade in the shape of GATT (now WTO) law. To this we can add fundamental rights in Kadi, ²⁹³ environmental protection in Intertanko ²⁹⁴ and Air Transport, ²⁹⁵ and consumer protection in ELFAA. These aims were achieved through autonomy (Kadi), direct effect (Intertanko), opportunistic interpretation of customary international law (Air Transport) and opportunistic interpretation of treaties (ELFAA). In each instance the Court is essentially balancing the substantive justice interest in respecting international law with other relevant substantive norms. But there has been a failure to consistently integrate substantive concerns into interpretative and other methods, meaning it is less than forthcoming while doing so and a clear picture for substantive justice fails to emerge in its reasoning. ²⁹⁶ In Simutenkov and Western Sahara, for example, we are less clear as to the norms at play, let alone their contours. Was the Court focused more on individual rights or the EU’s political relationship with Russia in Simutenkov? Was it institutional loyalty, a desire not to find the EU in breach of international law or another unexpressed factor that guided it in Western Sahara?

Ultimately there is little alignment of substantive justice with ‘thin’ justice so as to differentiate cases clearly. We have already identified multiple shortcomings in ‘thin’ justice previously and will develop further challenges concerning legal certainty below. These problems mean that significant pressure is being placed on substantive justice in this case law, pressure it may not be able to bear. The nature of substantive justice is such that case law cannot be confirmed as having failed in its pursuit of it, but the inverse is also true. It is a less than solid foundation from which to mount a defence of the case law for the Court, not least because a strong, consistent vision of substantive justice would be expected to trickle-down and create coherence in ‘thin’ justice. We will also see in subsequent developments focused on legal certainty that the Court’s ‘vision(s)’ of substantive justice have often not been shared by key stakeholders.

6. Legal certainty emphasis: further uncertainties

We have explored ‘thin’ justice and the visions held by the Court for substantive justice, exposing shortcomings in the former but necessarily only arguing that there are concerns for the latter. Of course, those developments have also given rise to significant legal uncertainty. However, what is remarkable is that fresh developments mean that for each of the three strands (bilateral treaties, multilateral treaties and treaties impacting third parties) there is further uncertainty on the horizon. Interventions from other institutions and criticism from scholarship and other courts/judicial bodies now raise pressing questions as to how the Court may respond.

A. Bilateral treaties

For new bilateral treaties there is an emerging practice of placing restrictions on direct effect either within agreements themselves or in Council decisions implementing such agreements. ²⁹⁷ There is also effort to make clear that VCLT interpretations are to be applied, not teleological ones. This is so in the EU–UK Trade and Cooperation Agreement, which also adds, no doubt with an eye to Polydor, ‘[f]or greater certainty, neither this Agreement nor any supplementing agreement establishes an obligation to interpret their provisions in accordance with the domestic law of either Party.’ ²⁹⁸

Such provisions have also recently been applied in practice. For instance, a panel of experts convened under the EU–South Korea Free Trade Agreement specifically applied VCLT interpretation, as was required by Article 14.16 of the Agreement. ²⁹⁹ The arbitral body also went into some detail in noting the ‘holistic’ nature of Article 31 VCLT, and observing ‘[a] disproportionate reliance on one particular element may yield a misplaced or inaccurate interpretation.’ ³⁰⁰ In emphasising interpretation was ‘all under the rubric of good faith’, no fewer than eleven cases were cited from the ICJ and WTO. ³⁰¹ International, ‘mindful’, ‘cautious’ and noting ‘the confines of Article 31’, ³⁰² this is a method which accentuates the most conservative of VCLT readings, and therefore contrasts significantly with the EU teleological approach. It may also contrast with the Court’s own approach to VCLT.

As such it becomes crucial to know which path will be taken by the Court when. But on this point there is a notable complication; will any attempted exclusion of its jurisdiction or direction concerning interpretation of international agreements be respected by the Court? ³⁰³ The prevailing view is that where such a restriction appears in the agreement itself ³⁰⁴ the Court would be bound, ³⁰⁵ but that where the restriction appears in implementing legislation it is more debatable. ³⁰⁶ Before turning to the latter case it is worth recording what is conceded by the author to be very much a minority view regarding the former.

It is submitted that it is hard to be so confident regarding the Court’s approach, especially in external relations; discretion has been retained in nearly all instances, and even where it appears to have run out it has often been resuscitated. ³⁰⁷ To be sure, the window to not follow a direction within the agreement itself is small, but it does appear to be there. The Court has previously stated ³⁰⁸ :

In conformity with the principles of public international law Community institutions which have power to negotiate and conclude an agreement with a non-member country are free to agree with that country what effect the provisions of the agreement are to have in the internal legal order of the contracting parties. Only if that question has not been settled by the agreement does it fall for decision by the courts having jurisdiction in the matter…

However, a lot of time has passed since this judgment in 1982 ³⁰⁹ and the discourse of both international and EU law on this point may have stagnated and shifted respectively.

At international level directions which are found by courts specifically to govern domestic application are both vanishingly rare and (increasingly) old. ³¹⁰ More recent ICJ case law would instead seem to suggest continued reluctance concerning such an obligation. ³¹¹ Nollkaemper goes as far as to say ‘[c]ompared to EU law, the support that international law provides for direct effect is exceedingly weak’, adding ‘[t]here is no authoritative judgment of an international court that direct effect is required’. ³¹² It makes little sense to think that international law should be stricter concerning a requirement expressed in a treaty for non-application of its rules.

Away from international law, if the Court were minded it could certainly draw on concepts within EU law to justify its continued involvement, and given the concept of autonomy this aspect may also ultimately be more important. ³¹³ International agreements are ‘an integral part’ ³¹⁴ of EU law, but they rank below EU Treaty provisions and ‘cannot affect the allocation of powers fixed by the Treaties’. ³¹⁵ The Court has previously held that even EU Treaty provisions cannot be given a ‘strict interpretation’ to limit the Court’s role, ³¹⁶ which is to ‘ensure that in the interpretation and application of the Treaties the law is observed’. ³¹⁷

In this light fresh direction in the Lisbon Treaty that the EU must inter alia contribute to ‘strict observance’ of international law may be enforceable, albeit so far its impact in case law appears mixed. ³¹⁸ There is also prominent place given to the concept of coherence in external relations, ³¹⁹ which could be impacted by allowing uneven judicial oversight of international agreements. Issues (somewhat ironically) of legal certainty/legitimate expectations ³²⁰ and equal treatment could also be operationalised by the Court to build a bridge between new international agreements and equivalent EU rights or existing jurisprudence in external relations case law. Perhaps partly for this reason non-discrimination clauses are conspicuously absent from some new Association Agreements. ³²¹

Emerging literature advocating (and occasionally evidencing) the external application of the EU Charter of Fundamental Rights ³²² could also be drawn upon; fundamental rights issues have been a solid starting point for the Court to gain a foothold in the Common Foreign and Security Policy (from which it is largely excluded). ³²³

The above appear to be at least viable arguments or avenues for future judicial manoeuvre for the Court to acquire a greater role where agreements seek to exclude it. This is not to say that any of these arguments are thoroughly convincing in limiting the effect of restrictions within agreements on domestic application, but just to note that they – and countless others no doubt – are there. Assumptions that a point is decided, and the Court will never shift, may be overvalued.

Concerning provisions in agreements on interpretation (rather than jurisdiction/direct effect), the assumption also seems to be that such direction will be followed. ³²⁴ Suffice to note, experience shows precise directions in the Swiss agreements and those concerning the European Economic Area appear not to have been followed (in the former the Court did not go as far as directed, in the latter it went further). ³²⁵ There is also scope – potentially – to reason that a direction to interpret under VCLT does not preclude teleology and may (depending on the nature of the agreement) require it. Bjorge suggests that even the Court’s (meta-)teleological approach towards EU law is wholly within the realms of VCLT; ‘it would be wrong to see the approach of the European Court of Justice as wholly out of touch with the general rule of interpretation, or as much more teleologic than that which follows from that approach’. ³²⁶ Once more, this is not an argument which necessarily persuades – it is disagreed with in this Article. But it is present, and to suggest alternatively that a provision in an agreement which mandates VCLT interpretation is ‘[s]uch a clear statement [that it] excludes any possible teleological interpretation by the CJEU’ ³²⁷ is not wholly persuasive either. Lines are blurred: whilst we have seen that the Court itself draws a distinction between VCLT and teleology, it has never indicated that the two approaches are in conflict. It seems open to judicial finesse that they are somehow complementary, that one follows or leads to the other. The Court has already conflated interpretative methods in some of its case law, which erodes further the extent to which directions on interpretation should be treated as decisive. ³²⁸

Concerning the binding effect of directions to the Court in implementing legislation, as compared to provisions in international agreements themselves, views differ more greatly. Such practice is relevant, inter alia, to the EU agreement with Korea ³²⁹ and to new Association Agreements with Ukraine, Moldova and Georgia. ³³⁰

Looking back to a previous manifestation of this practice, it is often noted that AG Saggio had no hesitation in stating, concerning a Council Decision implementing and seeking to restrict WTO law, ‘[a]lthough the wording of the recital is clear’ it is ‘simply a policy statement and, as such, cannot affect the jurisdiction of either Community or national courts to interpret and apply’ the agreement. ³³¹ However, ultimately the Court found that WTO law did not have direct effect. The Court referred to the Council Decision, but only after reaching its conclusion: ‘[t]hat interpretation corresponds, moreover, to what is stated in the final recital in the preamble to [the] Decision’. ³³² Depending on one’s view this rendered the Decision relevant or irrelevant to the outcome. At its most potent, though, directions in implementing decisions seem not to be more than a factor, rendering them far from decisive. ³³³

This uncertain status would seem to be reinforced by considering the Court’s varied approaches in a similar area of law. The ‘implementation’ and ‘reference’ principles, whereby secondary legislation may be adopted to achieve specific aims or indicate the intention to be bound by international law, appeared in the Nakajima ³³⁴ and Fediol ³³⁵ cases respectively. It is fair to say that the Court has since treated such directions as far from dipositive, and after oscillating case law ³³⁶ (including markedly divergent outcomes before the General Court and Court in one case) ³³⁷ the principles now seem limited to WTO law and probably only anti-dumping within that. ³³⁸ This means essentially that provisions in secondary legislation containing directions that ‘[t]his Directive shall apply, in accordance with international law’ ³³⁹ or that are ‘laying down rules to apply the provisions of the [Aarhus] Convention to Community institutions and bodies’ ³⁴⁰ are, after some confusion, ineffective. ³⁴¹

At the very least, early litigants concerning possible exclusions of direct effect and instructions on interpretation will face significant uncertainty. Given the Nakajima and Fediol sagas, this may become more prolonged. It is true that should direct effect be excluded then indirect effect would remain possible. However, aside from uncertainties associated with indirect effect, ³⁴² this approach would also inevitably lead to different and less favourable treatment compared to agreements with direct effect.

Accordingly, seeking to limit the Court’s role creates an awkward ‘thin’ (and arguably substantive) justice outcome whereby beneficiaries of established but less ambitious agreements may have wider judicial protection than those of newer and more ambitious ones. ³⁴³ Above all, it raises significant legal uncertainty, as the scope and effect of such provisions remain unclear. Whilst a Commission and Council initiative, the Court cannot be exempted from some responsibility for this state of affairs. First, much of the uncertainty now revolves around how the Court might respond, questions that simply would not arise with a more docile or deferential judicial body. Second, while rationales for refusing direct effect or excluding teleology in new agreements have been debated – including links to WTO law (which does not have direct effect) and preference for dispute settlement mechanisms – legal uncertainty through ‘surprises in judicial integration’ have also been mentioned as influencing this change in approach. ³⁴⁴

B. Multilateral treaties

Multilateral treaties are less likely to be fully aligned to EU values as, unlike bilateral treaty negotiations where the EU is often the stronger partner, the EU will be one of multiple participants ³⁴⁵ or merely succeed Member States. ³⁴⁶ Ultimately, the Court has frequently refused to apply multilateral treaties where they conflict with EU norms or interests (albeit not expressing this as the reason for refusal) ³⁴⁷ rather than expand the practice of strained interpretation exhibited in ELFAA. As discussed above, declining to apply multilateral treaties (application) is in principle more acceptable under international law than providing an incorrect interpretation. ³⁴⁸ Indeed, where these approaches have been challenged they have often concerned interpreting the ‘broad logic’, ³⁴⁹ ‘nature’ ³⁵⁰ or ‘spirit’ ³⁵¹ of an international agreement in order to refuse direct effect, methods which are increasingly noted to provide much discretion to the Court. ³⁵² There is certainly scope to argue that complexities in interpreting direct effect in landmark cases are in part the consequence of a relocation in the exercising of discretion, caused by pressure exerted on the Court’s interpretation of multilateral treaties.

Indeed, WTO law has often been at tension with EU interests. Having reconfirmed that its provisions are to be interpreted ‘in accordance with customary rules of interpretation of public international law’, ³⁵³ the Court has instead developed myriad approaches to overcome its limited discretion in interpretation. These include controlling when WTO law may be relied upon through refusing direct effect, applying reference and implementation exceptions, developing a ‘muted dialogue’ ³⁵⁴ with WTO Dispute Settlement Bodies and exploring potential EU liability to its own citizens for breaches of WTO law (which was entertained ³⁵⁵ but ultimately refused, ³⁵⁶ in spite of AG Maduro arriving at the converse conclusion through consideration of ‘the mainstream of the settled case-law’ ³⁵⁷ ).

Refusing direct effect in particular has been a method deployed by the Court in other multilateral case law. Concerning the Aarhus Convention the Court refused direct effect and unexpectedly declined to apply the Nakajima and Fediol exceptions; ³⁵⁸ this contrasted with the Advocate General’s close analysis of whether the relevant EU Regulation could be excluded from review not based on EU direct effect but on the basis of the Aarhus Convention’s own rules, applied according to VCLT rules of interpretation. ³⁵⁹ Intertanko, ³⁶⁰ refusing direct effect of UNCLOS, was criticised for finding that the international agreement’s nature was such that it could not confer rights on individuals. ³⁶¹ This was also in stark contrast to the Advocate General, who found no problem granting direct effect. ³⁶²

It is notable that the Court has shown a propensity to steer clear of directly interpreting multilateral provisions. As Mendez puts it, where ‘interpretation would appear to provide little shelter’ in allowing the Court to protect EU legislation – as in the ‘litmus test’ of Intertanko – ‘closing the gateway’ by refusing direct effect appears to be the Court’s solution. ³⁶³ Essentially, subsequent to ELFAA, refusal of direct effect, rather than interpretation, has frequently become the locus for exercising discretion in this area, with familiar criticisms for the impact this has on certainty inevitably tracking on the same point. ³⁶⁴ Interpretation of multilateral treaties, accordingly, has not generated the same level of controversy since. ³⁶⁵ But that does not mean that interpretation has not been a relevant, even decisive, factor guiding developments elsewhere.

Moving forward, there are also some signs that interpretation is becoming a key locus for exercising discretion once more. Diakite saw the Court define ‘internal armed conflict’ more broadly in an EU Directive than is provided for in international law under the Geneva Conventions, which require ‘organised armed groups… under responsible command’, control over territory and ‘sustained and concerted military operations’. The Court did not cite VCLT nor the familiar obligation to interpret EU legislation in light of international law. ³⁶⁶ Instead it adopted ‘usual meaning in everyday language of ‘internal armed conflict’’ defining it as where ‘a State’s armed forces confront one or more armed groups or in which two or more armed groups confront each other.’ ³⁶⁷ It did so by reasoning that the EU Directive was aimed at providing specific protection for the relevant applicant, which differed from the objectives of international law.

Kanavape concerned Cannabidoil (CBD), which is not known to have any ‘recognised psychoactive effects’ ³⁶⁸ but is derived from the cannabis plant. This, however, meant that under the Single Convention on Narcotic Drugs 1961 it was prohibited and failed to fall into narrow exceptions. The Court did not follow the Advocate General’s strained interpretative efforts to fit CBD within the exceptions. ³⁶⁹ Instead, the Court engaged directly with VCLT but claimed a ‘literal’ interpretation would not be appropriate as the preamble to the Convention focused on the health and welfare of mankind and commentaries on the Convention reveal that blanket inclusion of cannabis derivatives simply reflected the state of scientific knowledge at the time. This is clearly a questionable application of VCLT. The context purported to dislodge ‘ordinary meaning’ seems open to challenge through logical caution in approaching items with some harmless and some harmful properties and/or applications. It does not seem to be ‘manifestly absurd or unreasonable’ so as to justify recourse to supplementary means of interpretation under Article 32 VCLT. Ultimately the Court’s substantive outcome has been vindicated as subsequently CBD was de-listed from the international convention.

Legal uncertainty associated with the varied methods applied by the Court in these two cases has received little attention. Perhaps this is not surprising given the substantive justice outcomes pursued in Diakite and Kanavape – wider protection for refugees and de-listing of a substance apparently destined to be de-listed at international level anyway – can be seen as uncontroversial in scholars’ accepting flexibility in interpretation. ³⁷⁰

But if the anticipation is that the Court will only diverge from VCLT in limited circumstances that give rise to no controversy, it should be noted that ultimately the vote to de-list CBD at international level was decided by 27 States voting for, versus 25 against (with one abstention). ³⁷¹ At the time of judgment, then, the Court demonstrated that it had retained some risk-appetite for stretching VCLT interpretations: just a one vote swing and the substantive justice outcome could have given way to less favourable analysis of legal uncertainty in the method applied.

C. Treaties impacting on third parties

Criticism of the Court in ELFAA has been suggested to have caused (at least for some time) deviation from the path of opportunistic interpretation of international rules for multilateral agreements. The question concerning treaties impacting third parties becomes whether such a change may occur here, too? And, perhaps, why has change not occurred already?

Basedow suggests that constraints on the Court’s reasoning emerge from ‘interpretation techniques that delimit the range of legally admissible rulings’ ³⁷² in conjunction with the ‘varying preference constellations among Member States, European Commission and Parliament that delimit a subset of politically viable rulings’. ³⁷³ This may help to explain why case law in this area is emerging as problematic, given that views from Member States and EU institutions have been far from consistent concerning Morocco. ³⁷⁴ As such, in stretching international interpretations the Court is not facing the same ‘non-compliance or political override’ risk it may do should it contradict EU institutions or Member States en masse (for example concerning bilateral treaties, where as a counter to this it could seek to more closely rely on international law for legitimacy).

While the Court is not facing institutional challenge in the same way it has for bilateral treaty interpretation, the Court’s approach in this area has received especially negative attention elsewhere. In domestic proceedings reviewing the UK’s post-Brexit arrangements with Morocco (largely identical to the EU’s), a UK court was remarkably vocal on the matter. Concerning interpretation under Article 31 VCLT, Cockerill J noted that the EU Court’s approach ‘is not without its critics’ and ‘[t]here is no clear justification for the weight given to Article 31(3)(c) in the context of the rest of Article 31’. ³⁷⁵

Accordingly, the Court has found flexibility in interpretation where others find little. Of course, unlike the EU, needing to reason away tensions with international law, any potential invalidity had already been resolved by the UK Administrative Court through succinct reference to the UK’s dualist approach: ‘[a] decision of the Government to enter into a treaty is not reviewable by the Courts. It is non-justiciable.’ ³⁷⁶ There is some irony that the EU approach to interpretation of international law draws criticism from a court which itself is simultaneously declining to apply international law: a reminder once more of the contentious nature of interpretation compared to application.

Alas, interpretation is the path the Court has selected and Basedow’s rubric also takes into account the ‘risk of grave reputational damages’ as a further constraint on legal reasoning. ³⁷⁷ Pressure is undoubtedly increasing and is not limited to scholarship alone. ³⁷⁸ In light of increasing criticism it will now be interesting to see how the Court navigates when it is pushed further on this point once more. Odermatt observes at current ‘there would be very limited circumstances when the EU could ever violate international law by concluding an international agreement, perhaps only if it were manifestly unlawful, and the Court could not ‘interpret away’ the incompatibility.’ ³⁷⁹

We may soon find such a challenge for the Court. The cases currently on appeal concern whether the Council sufficiently engaged with the people of Western Sahara so as to gain their consent (under Article 34 of the VCLT in conjunction with the right to self-determination). ³⁸⁰ These were obligations emphasised by the Court itself in Western Sahara. The General Court considered that this obligation had not been fulfilled and annulled the relevant decisions, with delayed effect. Given that the General Court found ‘it is clear from the stipulations of the agreement at issue that the European Union and the Kingdom of Morocco expressed a common intention to apply that agreement to Western Sahara’, ³⁸¹ in this instance it seems harder for the Court to avoid conflict through interpretations presuming compliance with international law. But, inevitably, one cannot be so sure.

D. Appraisal

Uncertainty has flowed from shortcomings in distinguishing cases (‘thin’ justice) and challenges based on substantive justice (related to evolving EU interests in the face of international law and contentious substantive outcomes, exacerbated by the lack of clear engagement of such issues in the Court’s reasoning).

Beyond these problems this section has highlighted fresh, additional problems for legal certainty. Scholarship has been critical of the case law across all three strands of bilateral treaties, multilateral treaties and treaties impacting third parties. The Court had proved willing to listen to criticism in the multilateral sphere, albeit teleology may be re-emerging. It remains uncertain whether the Court will be responsive to criticism from scholarship and other courts concerning treaties impacting third parties. Somewhat ironically, in that strand of its case law the Court is demonstrating some deference and trust in other EU institutions. This is emphatically not reciprocated by the Commission and Council in the bilateral sphere, where there is effort to limit the Court’s role. The outcome of efforts to restrict the Court’s role through secondary legislation implementing agreements is far from certain. Views that statements even within international agreements will be decisive may also be misplaced, especially where they concern interpretation.

In the case law overall there does at least appear to be scope within the current balance for the Court to improve its delivery of legal certainty in individual cases through greater elaboration upon the selection and deployment of interpretative methods. This is noted to be all the more valuable where existing practices are modified. ³⁸² One might intuitively expect, even if the Court does not accept all efforts to limit its role, at least the prospect of significant uncertain teleological advances in the name of substantive justice to have lessened. Then again, if the direction of travel continues to curtail the Court’s influence it may not go quietly. Given that limits to teleology in bilateral agreements concerning residency are thought to be a concession based on politics, including ‘tribute to the Member States’ concerns about migration’, ³⁸³ one can note that the politics has emphatically changed. The Court is now less incentivised to keep to old accommodations and may become even more robust across the piece. In this regard it is notable that the Court is once more beginning to revisit teleology in the multilateral context and has so far been strong-minded in continuing its ‘unique’ approach to international rules for treaties impacting third parties.

Change, or anticipation thereof, looms large. In bilateral treaties there is effort to force change by other institutions, in multilateral treaty interpretation there may be emergent change initiated by the Court and in treaties impacting third parties scholarship (supported uniquely by a UK court) is pressuring for change. In short, there are new uncertainties in an area of law which can ill-afford them.

7. Conclusion

Given the contentious nature of interpretation, there does appear at least arguable discretion afforded from the international perspective for the Court’s varied approaches. This has remained a focus in scholarship and there is typically the suggestion (and concern) that the Court’s direction of travel is becoming less respectful of international standards of interpretation, ³⁸⁴ even if some scholars take a different view. ³⁸⁵ As a practical point, controlling international law’s domestic effects through rules governing application (such as direct effect, autonomy or even proportionality) ³⁸⁶ could be less contentious from the international perspective than those of interpretation. ³⁸⁷ This would reflect the fact that some view interpretation as being a shared enterprise (international-contested), whereas application is a matter for individual States (domestic-accepted). ³⁸⁸ This Article, though, has sought to highlight pressing concerns beyond the debate of fidelity to international law.

We have considered the EU’s approach to interpretation of bilateral treaties, multilateral treaties and treaties impacting third parties. The case law has been measured against the imperatives of justice and legal certainty, which have been argued to provide clearer benchmarks from which to draw conclusions than those focused on fidelity to international law. In balancing the need for justice and legal certainty the Court’s ‘flexible’ selection and deployment of VCLT and teleological interpretative methods at first glance prioritises justice over certainty. But on closer inspection, it may not satisfy either.

Whilst teleological interpretation has frequently been applied to bilateral treaties and has generally been expanding, the rationale for this has been unclear and is subject to incoherent exceptions. In determining interpretative method there is inconsistency in the relevance and weight afforded to; agreement type (Association, PCA, free trade etc.), treaty partner (close or more arms-length) and/or specific provisions on interpretation within agreements (eg Turkish compared to Swiss agreements). There is also absence of justification for applying divergent methods in apparently similar cases (eg Polydor and Commission v Italy). The Polydor test had long been vague and Simutenkov appeared to render it meaningless.

Beyond the bilateral context there are shortcomings in adherence to VCLT even where it is purported to be applied (ELFAA and case law on treaties impacting third parties). This is particularly problematic for legal certainty. In terms of selecting the relevant interpretative approach, whilst greater caution was evident in multilateral treaty interpretation subsequent to ELFAA, this is already subject to creeping exceptions. ³⁸⁹

Limited elaboration from the Court is the common contributory theme running through many of these shortcomings; but are these silences and ambiguities necessary in the name of justice? It seems doubtful. After all, silence and ambiguity impacts not only legal certainty but also raises questions concerning ‘thin’ justice, with unlikenesses between cases challenging to identify. Saying less provides greater freedom for a court in future cases, but it also weakens understanding and coherence of substantive justice conceptions. Arguably, if it is essential that the power of discretion remain with the Court in order to pursue substantive justice, the exact basis and rationale for variations in its exercise across the case law has been hard to identify and requires clearer communication (let alone acceptance).

In this light, it is worth noting that expansive interpretations of bilateral treaty provisions do not always sit comfortably with Member States, and scholars heavily criticised the Court’s interpretation of the multilateral Montreal Convention and its ongoing approach to treaties impacting on third parties. Perhaps most notably, the Commission and Council are actively seeking to limit the Court’s role in interpreting new bilateral agreements. All of this suggests that the substantive justice conceptions arrived at by the Court have been contentious. Given that legal certainty and ‘thin’ justice are also problematic, this renders the Court’s position rather precarious.

The path forward has also become increasingly complex. For example, while closer adherence to international interpretative standards may discourage the Commission and Council from seeking to exclude the Court from bilateral agreements, and could dampen criticism of multilateral case law and treaties impacting third parties, it would necessitate change in the Court’s approach. This would impact further on ‘thin’ justice as new cases would be treated differently from old. This would also, inevitably, call further into question the existing substantive justice merits of this area; an issue on which (even if unsatisfactorily articulated) the Court itself clearly places great significance. Indeed, the final ‘moonshot’ for justification on which much case law seems to rest is substantive justice. From this perspective there may be some disinclination from the Court to change course. With divergence a significant chunk of case law would have no merit left to it. It could be labelled uncertain and unjust in the formal and substantive sense; the unwanted trinity which marks failure. ELFAA and Simutenkov come dangerously close to achieving this status already.

If, in changing approach, it is instead argued simply that the Court is now adhering to instructions from EU institutions regarding its role and approach to interpretation (an arguable ‘unlikeness’ compared to previous case law), why does it not follow such instructions in other areas (eg with the Nakajima and Fediol exceptions beyond WTO law)? Ultimately, in practice, would acceptance of such a public rebuking from other EU institutions not be an unimaginable low-water mark for a court that once told the EU and the world that a ‘new legal order’ ³⁹⁰ had been established and later (essentially) the UN Security Council that it had to think again? ³⁹¹ There are also questions to be answered elsewhere. Can the Court really accept that it has got it wrong concerning treaties impacting third parties? Will it resist the temptation to bend VCLT to suit its own purposes in multilateral case law? How will the Court manage these challenges and what precedent may failing to manage them effectively set concerning the role and credibility of the Court in other integral parts of EU law?

Taking stock, the Court is in a remarkably awkward position. It is submitted that this problem is based essentially on shortcomings in ‘thin’ justice, substantive (‘thick’) justice and legal certainty. If stakeholders do not consider substantive justice to be done, then most courts can appeal to serviceable legal certainty and ‘thin’ justice (in treating like cases alike) as mitigation. Looking back, in this area of its case law, the Court often cannot do so. Looking forward, balancing these imperatives appears to be getting harder, with new complexities and a weaker than neutral starting point.

For the Court and others, there are lessons to be drawn from shortcomings concerning justice and certainty in this area. First, not to pay proper attention to legal certainty is problematic. Second, it is especially foolish not to do so in those instances in which it is not operating as a specific cap on justice (ie where a balance between the two aims is not required to achieve the outcome sought). ³⁹² Third, if the justice pursued in a case is substantive then any court is on more shaky ground because substantive justice is inherently debatable in all but the most obvious circumstances. Fourth, pursuit of substantive justice becomes especially problematic if a court is not consistently administering workable levels of ‘thin’ justice and certainty in its surrounding case law: at this point a court becomes vulnerable as just a further voice in a political debate, rather than a body which resolves disputes through application of law. Fifth, if problems persist (uncertainty, not treating like cases alike and adoption of contested substantive justice outcomes) then actors may start to look elsewhere for means to resolve disputes, excluding courts. There will be a sixth lesson. It is due to begin soon and will concern, for better or worse, the theme of recovery.