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.Footnote 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.Footnote ²⁹ Other scholars have variously identified the ‘dual nature of law’;Footnote ³⁰ that ‘the law should strive to balance certainty and reliability against flexibility’;Footnote ³¹ that these are the ‘two conflicting and yet equally important objectives in any legal order’;Footnote ³² the need ‘to guarantee simultaneously the certainty of law and its rightness’;Footnote ³³ for ‘[t]he law must be certain. Yes, as certain as may be. But it must be just too.’Footnote 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.’Footnote 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.Footnote ³⁶

In stark contrast, Weber advocates the merits of ‘formal rational law’ (through legislation) supported by ‘formal rational administration of justice’.Footnote 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.’Footnote ³⁸

We could add other academics in between. Radbruch, having initially prioritised certainty even more greatly,Footnote 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.’ Footnote ⁴⁰ Alexy aligns himself closely with Radbruch’s model,Footnote ⁴¹ whilst Cardozo cites Munroe Smith with approval. Smith suggests thatFootnote ⁴² :

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’.Footnote 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’.Footnote 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.’Footnote 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’.Footnote 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,Footnote 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.Footnote 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.Footnote ⁴⁹ For these reasons any hostility towards international law often feels less acceptable coming from EU courts.Footnote 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,Footnote 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’Footnote ⁵² and, more robustly, that it is ‘a municipal court, a court of the European Union, not a general international court’.Footnote ⁵³ Von Bogdandy’s reasons for treatment of the Court as ‘domestic’ are especially persuasive:Footnote ⁵⁴

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).Footnote 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.Footnote 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.Footnote 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’.Footnote 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.Footnote 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’.Footnote ⁶⁰ When dealing with international law the scope of the relevant community to be considered in this regard broadens,Footnote ⁶¹ therefore Halberstam observesFootnote ⁶² :

But for othersFootnote ⁶³ :

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.’Footnote 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.’Footnote 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’Footnote ⁶⁶ ), equally, those seeking to impose greater rigour concede that inevitably some discretion remains and outcomes can differ.Footnote 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.Footnote 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.Footnote 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.’Footnote 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’.Footnote 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.Footnote 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,Footnote 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.Footnote ⁷⁴

However, due to ‘thick’ (or substantive) justice’s contested nature – Nicol suggests we ‘celebrate justice as something we argue about’Footnote 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.’Footnote ⁷⁶

‘Thin’ (or formal) justice, however, is widely accepted. ‘Thin’ justiceFootnote ⁷⁷ amounts to the assertion that like cases should be treated alike and unlike cases unlike in proportion to their unlikeness.Footnote 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’Footnote 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.’Footnote 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.Footnote 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.’Footnote 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.Footnote 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.Footnote 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.Footnote 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).Footnote 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 greatlyFootnote 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.Footnote ⁸⁸

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

It went on to say that ‘contrary to what has previously been decided’ in Dassonville Footnote 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’.Footnote 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,Footnote 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.

A. Bilateral treaties

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.Footnote ¹⁹² This entrenches VCLT and makes attempts to move away from it often appear less than bona fide.Footnote ¹⁹³ However, extension of internal teleological interpretation to multilateral treaties has at times featured, starting with the European Economic Area Agreement (EEA).Footnote 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’Footnote ¹⁹⁵ it would be necessary to apply Article 31 VCLT to it.Footnote 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’Footnote 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.Footnote ¹⁹⁸ The new agreement continued to seek ‘as much homogeneity as possible’ but stopped short of creating a combined court.Footnote 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).Footnote 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.Footnote 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’,Footnote ²⁰² it appears subsequently to have gone beyond the Article 6 EEA requirement to only apply judgments ‘prior to the date of signature’.Footnote ²⁰³

This impacts upon EFTA members who ‘did not wish to commit themselves to future, unforeseeable developments in the case law’Footnote 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’Footnote ²⁰⁵ and, given the strong emphasis on homogeneity in the EEA,Footnote ²⁰⁶ some argue ‘there would have to be strong reasons for not extending EU law interpretations.’Footnote 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]’.Footnote ²⁰⁸ Over a decade on Ghazaryan writesFootnote ²⁰⁹ :

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.’Footnote 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,Footnote ²¹¹ a reluctance to advance teleological interpretation to rights related to residenceFootnote 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’Footnote ²¹³ for the purposes of ‘thin’ justice, the aim that ‘individuals may ascertain unequivocally what their rights and obligations’ cannot be met.Footnote 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.Footnote 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.

A. Bilateral treaties

Scholars have noted that (Switzerland asideFootnote ²¹⁶ ) a broad trend of extending internal interpretations to international agreements has continuedFootnote ²¹⁷ and Simutenkov Footnote ²¹⁸ arguably marks its peak. The case concerned a Partnership and Cooperation Agreement (PCA) with Russia.Footnote 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 caseFootnote ²²⁰ ), being described as an ‘entry-level agreement’,Footnote ²²¹ ‘on the whole rather meagre in terms of vision and concrete content.’Footnote 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.Footnote ²²³ Concerning the Russian agreement, Koutrakos draws particular attention to the emphasis on ‘mutual advantage, mutual responsibility and mutual support’ in substantive areas.Footnote ²²⁴ In the agreement with Morocco there was instead asymmetry,Footnote 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’.Footnote ²²⁶ The Court was facilitated in its reasoning due to determining the effects of a very similar provision in the Association Agreement with Slovakia.Footnote 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,Footnote ²²⁸ with the justification and capacity to anticipate application of the teleological approach diluted with each extension.Footnote ²²⁹

The outcome of Bosman is well known; sport can amount to an economic activity and be subject to EU law.Footnote 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’Footnote 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’.Footnote 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.Footnote 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’Footnote ²³⁴ due to ‘the aims and context of the Association Agreement’.Footnote ²³⁵ The Court expressly laid out the objectives of the Agreement asFootnote ²³⁶ :

The result was that the reasoning from Bosman could be fully ‘transposed’ onto Article 38(1) of the Association Agreement.Footnote 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.Footnote ²³⁸ This preference for redeploying matters decided in a previous case, in a very different context,Footnote 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).Footnote ²⁴⁰

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

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’.Footnote 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.’Footnote ²⁴³ Accordingly, he emphasises that PCAs ‘are a far cry from being an alternative to the [Association Agreements]’Footnote ²⁴⁴ and attempts by the Court to treat them as equivalent are noted to be ‘remarkable’.Footnote 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.Footnote ²⁴⁶ The EU’s political relationship with Russia has variously been cautious, co-operative, competitive or mired in crisis.Footnote 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,Footnote 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.Footnote 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.’Footnote ²⁵⁰

The Court acknowledged that the Convention limited liability of carriers to ‘4 150 [Special Drawing Rights (SDRs)] for each passenger’.Footnote 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.’Footnote ²⁵² The challenge was unfounded; the EU legislature was free to impose liability on carriers.

The Regulation coveredFootnote ²⁵³ :

The Convention, the Court argued, covered individual damage sustained by passengers. The judgment implicitly acknowledges the potential for some overlap with the ConventionFootnote 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).Footnote 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.Footnote 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,Footnote ²⁵⁷ as has the House of Lords (now UK Supreme Court).Footnote 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 lawFootnote 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.Footnote 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 Footnote ²⁶¹ where, in spite of Regulation 261/2004Footnote 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.Footnote 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.’Footnote 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 endorsedFootnote ²⁶⁵ :

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 judgmentsFootnote 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,Footnote ²⁶⁷ 2015,Footnote ²⁶⁸ 2016,Footnote ²⁶⁹ 2018,Footnote ²⁷⁰ and 2021 (featuring two judgements on a single day, Footnote ²⁷¹ both of which are currently under appealFootnote 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.Footnote 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)Footnote 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.Footnote 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).Footnote 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’.Footnote 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.Footnote ²⁷⁸ Indeed, further evidence that the Council and Commission intended to cover the contested territory was ignored by the Court.Footnote 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.Footnote 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:Footnote ²⁸¹

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.’Footnote 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’Footnote 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’.Footnote 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.Footnote 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.Footnote 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’.Footnote 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.Footnote ²⁸⁸ Internally proportionality has been a helpful tool in the Court’s incremental management of these aims.Footnote ²⁸⁹

In case law, applying international law proportionality is largely absent.Footnote 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’.Footnote 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.Footnote 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,Footnote ²⁹³ environmental protection in Intertanko Footnote ²⁹⁴ and Air Transport,Footnote 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.Footnote 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.

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.Footnote 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.’Footnote 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.Footnote 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.’Footnote ³⁰⁰ In emphasising interpretation was ‘all under the rubric of good faith’, no fewer than eleven cases were cited from the ICJ and WTO.Footnote ³⁰¹ International, ‘mindful’, ‘cautious’ and noting ‘the confines of Article 31’,Footnote 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?Footnote ³⁰³ The prevailing view is that where such a restriction appears in the agreement itselfFootnote ³⁰⁴ the Court would be bound,Footnote ³⁰⁵ but that where the restriction appears in implementing legislation it is more debatable.Footnote 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.Footnote 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 statedFootnote ³⁰⁸ :

However, a lot of time has passed since this judgment in 1982Footnote 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.Footnote ³¹⁰ More recent ICJ case law would instead seem to suggest continued reluctance concerning such an obligation.Footnote 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’.Footnote 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.Footnote ³¹³ International agreements are ‘an integral part’Footnote ³¹⁴ of EU law, but they rank below EU Treaty provisions and ‘cannot affect the allocation of powers fixed by the Treaties’.Footnote ³¹⁵ The Court has previously held that even EU Treaty provisions cannot be given a ‘strict interpretation’ to limit the Court’s role,Footnote ³¹⁶ which is to ‘ensure that in the interpretation and application of the Treaties the law is observed’.Footnote 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.Footnote ³¹⁸ There is also prominent place given to the concept of coherence in external relations,Footnote ³¹⁹ which could be impacted by allowing uneven judicial oversight of international agreements. Issues (somewhat ironically) of legal certainty/legitimate expectationsFootnote 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.Footnote ³²¹

Emerging literature advocating (and occasionally evidencing) the external application of the EU Charter of Fundamental RightsFootnote 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).Footnote 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.Footnote 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).Footnote 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’.Footnote 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’Footnote 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.Footnote 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 KoreaFootnote ³²⁹ and to new Association Agreements with Ukraine, Moldova and Georgia.Footnote 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.Footnote 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’.Footnote 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.Footnote 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 Footnote ³³⁴ and Fediol Footnote ³³⁵ cases respectively. It is fair to say that the Court has since treated such directions as far from dipositive, and after oscillating case lawFootnote ³³⁶ (including markedly divergent outcomes before the General Court and Court in one case)Footnote ³³⁷ the principles now seem limited to WTO law and probably only anti-dumping within that.Footnote ³³⁸ This means essentially that provisions in secondary legislation containing directions that ‘[t]his Directive shall apply, in accordance with international law’Footnote ³³⁹ or that are ‘laying down rules to apply the provisions of the [Aarhus] Convention to Community institutions and bodies’Footnote ³⁴⁰ are, after some confusion, ineffective.Footnote 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,Footnote 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.Footnote 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.Footnote 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 participantsFootnote ³⁴⁵ or merely succeed Member States.Footnote 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)Footnote 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.Footnote ³⁴⁸ Indeed, where these approaches have been challenged they have often concerned interpreting the ‘broad logic’,Footnote ³⁴⁹ ‘nature’Footnote ³⁵⁰ or ‘spirit’Footnote ³⁵¹ of an international agreement in order to refuse direct effect, methods which are increasingly noted to provide much discretion to the Court.Footnote 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’,Footnote 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’Footnote ³⁵⁴ with WTO Dispute Settlement Bodies and exploring potential EU liability to its own citizens for breaches of WTO law (which was entertainedFootnote ³⁵⁵ but ultimately refused,Footnote ³⁵⁶ in spite of AG Maduro arriving at the converse conclusion through consideration of ‘the mainstream of the settled case-law’Footnote 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;Footnote 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.Footnote ³⁵⁹ Intertanko,Footnote 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.Footnote ³⁶¹ This was also in stark contrast to the Advocate General, who found no problem granting direct effect.Footnote 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.Footnote 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.Footnote ³⁶⁴ Interpretation of multilateral treaties, accordingly, has not generated the same level of controversy since.Footnote 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.Footnote 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.’Footnote 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’Footnote 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.Footnote 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.Footnote 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).Footnote 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’Footnote ³⁷² in conjunction with the ‘varying preference constellations among Member States, European Commission and Parliament that delimit a subset of politically viable rulings’.Footnote 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.Footnote 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’.Footnote 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.’Footnote 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.Footnote ³⁷⁷ Pressure is undoubtedly increasing and is not limited to scholarship alone.Footnote 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.’Footnote 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).Footnote 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’,Footnote 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.Footnote 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’,Footnote 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.