Constitutionalising the end of history? Pitfalls of a non-regression principle for Article 2 TEU

Abstract

Enforcement and conceptualisation of Article 2 TEU values – Rule of law – A non-regression principle for EU values faces significant pitfalls – Limits of the ‘backsliding’ paradigm informing doctrinal developments under Article 2 TEU – Inadequacy of a progress/regression trajectory as a lens for constitutional developments – Complexity of Article 2 values threatens to render regression assessments simplistic – Non-regression and the equality of member states – Potential conflicts between non-regression and minimum standards as tests for Article 2 compliance

Introduction

In recent years, the scope and manner of judicialisation of the values expressed in Article 2 TEU have become the subject of intense scholarly attention. ¹ At least since the groundbreaking ASJP case, ² the Court has proactively made use of Article 2 values in its jurisprudence, even referring to the values in Article 2 as making up the ‘identity’ of the EU. ³

Of the wave of Article 2-related case law passed down in recent years, the Repubblika case has attracted particular attention as a potential landmark in the ‘big picture’ development of limits on member states’ constitutional law based on the Union’s shared values. ⁴ When the European Court of Justice passed down its judgment in Repubblika, commentators noted that the Court had seemingly smuggled a new jurisprudential ‘hook’ for the judicial enforcement of EU values into an otherwise innocuous-looking judgment. ⁵ In short, the Court drew a connection between Article 49 and Article 2 TEU, arguing that the commitment of member states to Article 2 values is a condition for accession and therefore remains a condition during membership. From this, it followed that any ‘regression’ of a member state’s laws regarding these values is precluded by EU law. According to these commentators, the Court formulated what has been described as a ‘principle of non-regression’ for the values contained in Article 2 TEU.

What exactly the non-regression principle will come to mean remains shrouded in ambiguity. However, commentators were quick to note the potential of ‘non-regression’ as a milestone in the European Court of Justice’s quest to juridically flesh out the EU’s founding values in Article 2 TEU. They have hailed it as a bold step forward in the protection of EU values – rather than stipulating concrete requirements regarding the rule of law or other Article 2 values, the ‘non-regression principle’ would simply hold that any deterioration in the way member states give expression to these values will be incompatible with Article 2 TEU. Leloup, Kochenov, and Dimitrovs see in Repubblika a ‘grand opening, marking something new and potentially truly far-reaching’. ⁶ The three authors argued that ‘non-regression’ could become a key to solving the ‘Copenhagen dilemma’: whereas accession to the EU was supposed to spell the ‘end of history’ for Central and East European states transitioning to democracy – consolidating their status as liberal democracies – the EU’s limited ability to influence constitutional development of its member states, once acceded, is thought to have facilitated democratic backsliding in some of these states. ⁷ A non-regression principle, ambitiously extended to all the values mentioned in Article 2 TEU, rather than just the rule of law, could ostensibly spell an end to this dilemma. ⁸

In this paper, I will argue that a veritable non-regression principle for EU values would face significant problems and pitfalls that should render us sceptical of its merits. Maximalist approaches to non-regression sound extraordinarily compelling – a way of unfolding the full potential of the shared values of the Union. However, the very idea of non-regression risks obscuring the deep complexities that come with the territory of Article 2 TEU and threatens to wrongly reduce legislative and constitutional changes to a simplistic trajectory of progress and ‘backsliding’. Further to that, non-regression also raises significant questions regarding the equality of member states. Three potential pitfalls will be discussed in-depth.

First, applying a ‘regression’ lens to Article 2 threatens to reduce constitutional developments and changes to an inadequate linear trajectory, allowing for findings of either progress or ‘regression’ in the realisation of Article 2 values. The ‘backsliding’ paradigm implicit in non-regression certainly has its place, but it should not induce us to overlook more complex and nuanced constitutional developments that cannot be placed on such a linear trajectory. Otherwise, non-regression runs the risk of giving way to capricious enforcement of Article 2 requirements: it might lead us to substitute EU legal requirements for a member state’s own capacity to constitutionally regenerate from illiberal governance, or even stifle good faith attempts at constitutional innovation or experimentation for fear of ‘backsliding’.

Second, ‘non-regression’ runs the risk of eschewing the complexity of the values in Article 2 TEU as well as the ways in which member states’ respective constitutional developments give expression to them. The difficulty of conclusively identifying certain changes as ‘regressive’ must not be understated: not all ‘steps back’ pertaining to safeguards for certain values necessarily endanger those values.

Third, the explicit link between non-regression and accession to the EU threatens to exacerbate the inequality of member states by establishing double standards regarding their measure of constitutional autonomy. Standards of ‘regression’ will vary among member states, depending on the laws currently in force, or the laws in force at the time of accession. Non-regression is thus bound to give rise to an uneven and asymmetrical enforcement of Article 2 TEU. More than that, non-regression runs the risk of reproducing the power asymmetries between those member states that acceded before the Copenhagen criteria were systematically monitored in the accession process and those that were subject to extensive accession conditionalities.

To some extent, these pitfalls also apply to less ambitious conceptions of non-regression that seek to limit its use to ‘significant’ regressions only. A ‘minimalist’ conception that merely regards non-regression as an anchor for concrete minimum requirements can hardly be understood as a non-regression principle to start with. Finally, it will be argued that the concrete place and role that non-regression will take within the Court’s case law remain unclear. The Court might face difficulties in reconciling its non-regression case law with other existing minimum requirements or ‘red lines’ posed by Article 2 TEU.

This article will first outline the Court’s formulation of the principle of non-regression in Repubblika before placing the idea of non-regression in context with other examples of non-regression principles in international and European law. The following sections will then discuss the three pitfalls of non-regression outlined above. The final two sections will explore the potential for more ‘minimalist’ versions of non-regression and raise questions about non-regression’s ‘fit’ within the existing case law. The conclusion will summarise the arguments raised and draw wider implications for the judicialisation of Article 2 TEU.

The non-regression principle in Repubblika and beyond

The Court first formulated a principle of non-regression in the Repubblika case, ⁹ which concerned the reform of judicial appointment procedures in Malta. Taking note of recent developments in its jurisprudence on judicial independence, ‘Repubblika’, a Maltese civil rights non-governmental organisation, brought an actio popularis before the Maltese Constitutional Court, arguing that the judicial appointment procedures as laid down in the Maltese Constitution are contrary to EU law – specifically, Article 19(1) TEU and Article 47 of the Charter. The judicial appointment procedure, as provided for by the Maltese Constitution, would give too much discretion to the Prime Minister, raising doubts as to the independence of the appointed judges. ¹⁰ The question of the judicial appointment procedure’s conformity was thus referred to the European Court of Justice.

The Court, having regard to its already extensive case law on the question of standards for judicial independence, found no issue in the case at hand. Even though the procedure under the Maltese Constitution was one that gave considerable leeway to the Prime Minister to appoint candidates, the Court found that ‘that power is not such as to give rise to legitimate doubts concerning the independence of candidates selected’. ¹¹

In arriving at this conclusion, the Court took an unusual detour. Before applying the ‘appearances’ test for judicial independence that the Court had developed in previous cases, ¹² the Court embarked on a number of seemingly unrelated observations: the Court, first, noted that Malta had acceded to the EU under Article 49 TEU on the basis of a different constitutional provision, which did not provide for a judicial council to be involved in the appointment process. The Court then noted that, under Article 49, the Union is ‘composed of States which have freely and voluntarily committed themselves to the common values referred to in Article 2 TEU’. ¹³

This ‘voluntary commitment’ of the member states to the founding values of the EU under Article 49 implies that member states cannot step back from this commitment. ¹⁴ Accordingly, the Court found that ‘a Member State cannot therefore amend its legislation in such a way as to bring about a reduction in the protection of the value of the rule of law’. ¹⁵ Concretely, this means that ‘any regression of their laws on the organisation of justice is prevented, by refraining from adopting rules which would undermine the independence of the judiciary’. ¹⁶

Rather than effectively assessing the judicial appointment rules with reference to certain minimum requirements, the Court compared them to the provisions on the basis of which Malta acceded to the EU, considering whether the amended provisions would amount to a regression of Malta’s laws on the organisation of justice. The Court took note of the fact that the previous constitutional provisions came with fewer constitutional safeguards: whereas the current constitutional provision, as amended in 2016, involves a judicial council in the nomination process, the previous provision did not. ¹⁷ The Court then proceeded to examine the rules in more detail, finding that the rules surrounding the Judicial Appointments Committee seemed to guarantee its independence, that little doubt had been expressed by Maltese courts as to that Committee’s independence, and that the Prime Minister’s appointment powers are circumscribed by rules regarding necessary qualifications for candidates and a duty to give reasons. Accordingly, the Court concluded that the rules in question ‘do not appear [to] give rise to legitimate doubts, in the minds of individuals, as to the imperviousness of appointed members of the judiciary to external factors’, finally recalling the ‘appearances test’ the Court first formulated in AK. ¹⁸

The connection between Articles 2 and 49 TEU that stands at the core of the non-regression principle is not a novelty. The Court had taken note of this connection already in the Wightman case, ¹⁹ from which the link between the two articles gradually started percolating through the rule-of-law case law. ²⁰ This marked a subtle shift in how the Court of Justice makes the values in Article 2 TEU actionable. Prior to Wightman, the Court tended to place the primary emphasis on the functional importance of the ‘fundamental premiss’ of shared values to mutual trust between the member states. ²¹ But in the aftermath of Wightman, the Court started supplementing the importance of mutual trust with a logic of constitutional pre-commitment: ²² values are enforced not merely because they are necessary or functional, but because the member states committed themselves to them when they joined the EU.

What does constitute a novelty in Repubblika is that the Court delineated the consequences of this logic of pre-commitment – namely, that ‘compliance by a member state with the values enshrined in Article 2 TEU is a condition for the enjoyment of all of the rights deriving from the application of the Treaties’. ²³ Repubblika, in the eyes of many, has thus established a new basis for the European Court of Justice’s value jurisprudence, grounding it in a principle of non-regression that prohibits member states from ‘reducing’ the level of protection of the rule of law they committed to when they joined the Union. Since the member states have committed themselves to the values in Article 2 TEU upon acceding to the EU, they are under an obligation not to violate that commitment by reducing the extent to which those values are protected.

At the time of writing, the Court has explicitly made use of this principle on two occasions following Repubblika: first, in Commission v Poland (Régime disciplinaire des juges), the Court found that the introduction of the now-defunct Disciplinary Chamber of the Polish Supreme Court constituted a regression, reducing the protection of the rule of law in Poland. ²⁴ Second, in AFJR, the Court briefly invoked non-regression to justify the binding nature of the CVM Decision that enables continued monitoring of the rule of law and anticorruption under the Cooperation and Verification Mechanism. ²⁵

Non-regression in context

In both cases, the application of non-regression has been rather shallow in comparison to the methodology the Court had apparently outlined for non-regression in Repubblika. In Commission v Poland, the non-regression principle played second fiddle to the more substantive ‘appearances test’ for judicial independence; in AFJR, it did not find concrete application but was merely invoked to establish the binding nature of the CVM Decision. As a result, it has not become any clearer what shape exactly this principle of non-regression will take. Repubblika tells us little about the baseline from which a regression would be judged – would the baseline of non-regression align with the laws in place at the time the member state acceded, or with those in place before the change in question was enacted? Is there a de minimis threshold that needs to be crossed for a reduction in the protection of an Article 2 value to count as a ‘regression’? Or can a ‘regression’ only be found if certain minimum standards are being violated?

Non-regression clauses in other regulatory contexts

Principles of non-regression, of course, are nothing new. They exist in other contexts but are much more clearly defined, constrained by, and embedded in their distinct regulatory environment. Most prominently, perhaps, many bilateral and multilateral trade agreements and investment treaties feature ‘non-regression’ clauses especially in the field of environmental protection. ²⁶ For instance, Article 24.5 of the Canada–EU Trade Agreement states that ‘The Parties recognise that it is inappropriate to encourage trade or investment by weakening or reducing the levels of protection afforded in their environmental law’. ²⁷ The EU–UK Trade and Cooperation Agreement precludes reducing the level of protection provided by environmental and social standards ‘in a manner capable of affecting trade or investment between the Parties’, as part of the broader level playing field provisions that seek to prevent a regulatory race to the bottom. ²⁸ Other examples of non-regression clauses can be found in EU law itself – for instance, where directives stipulate that the minimum standards established by them shall not give cause to a member state to reduce an existing level of protection. ²⁹ International human rights law is similarly familiar with a principle of ‘non-retrogression’ in the area of socio-economic rights as a corollary of the broader duty of ‘progressive realisation’ of the rights contained in the International Covenant on Economic, Social, and Cultural Rights. ³⁰

Two crucial questions can be posed with respect to any of these non-regression clauses, and, by extension, to any possible non-regression principle for EU values. The first question pertains to the baseline from which a non-regression principle would depart to find a regression. This baseline can be static, taking a fixed point in time as a point of departure; or it can be dynamic, taking the most recent status quo as the point from which regression is assessed. In the case of the EU–UK Trade and Cooperation Agreement, a static baseline was established – the Agreement prohibits reduction of the levels of environment and labour protection ‘below the levels in place at the end of the transition period’. ³¹ Non-regression thus takes a fixed point in time as the place from which regression is assessed. While any new law must not reduce protection of a given value compared to the level of protection at the baseline date, it need not necessarily prevent regression from the laws currently in force. This seems to be different in the case of the Canada–EU Trade Agreement, where no such point in time is defined in the text. ³² It also mirrors the principle of ‘non-retrogression’ of economic and social rights, which is linked to the idea of ‘progressive realisation’ of such rights. Under a dynamic baseline, any new law must improve – or at least not reduce – protection compared to the old law.

The second question pertains to the threshold of regression. When is a change in law or practice to be considered a regression? The actual ‘measurement’ of regression will, of course, depend on the subject matter to which the clause in question pertains. However, non-regression clauses can, and often do, provide further indication as to what is necessary for a measure to qualify as ‘regression’. The EU–UK Trade and Cooperation Agreement features the qualifier ‘in a manner capable of affecting trade or investment between the Parties’, ³³ significantly narrowing the scope of non-regression. Reductions in the level of protection that do not affect trade or investment, either because the legislative change in question is immaterial to both, or because the reduction of protection in question does not lead to a conflict with the standards the other party has in place, would likely not be covered by the non-regression clause. The extent to which trade or investment are affected thus inadvertently also becomes the ‘scale’ against which regression is measured. Similarly, the Canada–EU Trade Agreement speaks of ‘encourag[ing] investment by weakening or lowering levels of protection’. ³⁴ In that case, the question is rather one of intent, independent of any actual effects. ³⁵ Non-regression rarely comes on its own terms and is rarely self-explanatory.

Imagining non-regression for Article 2 TEU

This brings us to the potential shape of a non-regression principle for EU values. Non-regression has enthralled the imagination of many legal scholars, leading some to regard non-regression as the key to extending the Court’s value jurisprudence beyond effective judicial protection to all values contained in Article 2. ³⁶ In this vein, many commentators have been keen to look at Repubblika in a maximalist way – that is, as establishing a veritable non-regression principle; even a form of ‘one-way ratchet’ that captures any regression in the protection of Article 2 TEU values. ³⁷ Leloup, Kochenov and Dimitrovs suggest that non-regression could function as ‘a ratchet, which does not allow member states to reverse the progress they have made’. ³⁸ Mader more explicitly expresses the view that the Court quite deliberately did not set a minimum threshold for what constitutes a regression. As Mader argues, ‘the ECJ’s regression prohibition should not be understood as implying a de minimis threshold for violations of the rule of law, but rather that any reduction in the rule of law […] is to be avoided’. ³⁹ Both see considerable value in such a maximalist non-regression principle – either as a means of judicially fostering member state progress in the realisation of these values, or as a safeguard against a slow rollback of legislation and constitutional provisions that protect these values.

Alternatively, one might conceive of non-regression in a more restrained fashion. Spieker tentatively understands the non-regression principle as only applying to significant regressions – given the fundamental nature of Article 2 TEU, a particular threshold of severity is implied. Accordingly, only ‘significant regressions from pre-existing national standards’ protecting Article 2 values should fall within the scope of the principle. ⁴⁰

A final alternative could be a very minimalist version of non-regression, according to which merely regressions beyond a particular level of protection would constitute a violation of Article 2 TEU. The Court’s reasoning in Repubblika may be ambiguous enough to allow such an interpretation: if one places importance on the concrete prohibition that followed from non-regression in Repubblika – namely, that member states must not adopt rules which would ‘undermine the independence of the judiciary’ ⁴¹ – this may be understood as establishing a binary test rather than a ban on regressions. The ‘non-regression principle’, then, may simply be the anchor on the basis of which more concrete tests, prohibitions and minimum requirements can be established. ⁴²

Non-regression and the limits of the backsliding paradigm

We can now turn to the three pitfalls of non-regression that this article set out to describe. The first of these pitfalls pertains to the conceptions of progress and regression underlying more ambitious approaches to non-regression, and, in particular, the ‘democratic/constitutional backsliding’ paradigm that seemingly informs non-regression. Applying the idea of non-regression to EU values seems to presume that the adherence of member states to Article 2 TEU values could be measured along a linear trajectory, allowing for findings of progress and regression. At least tacitly, the idea of non-regression seems to translate this idea of a trajectory of progress into law. Some institutional arrangements can signify progress towards the values encapsulated by Article 2 TEU, while others are ‘regressive’ and undo progress already made. The non-regression principle thus seems to provide a route from merely ensuring a ‘baseline’ compliance with EU values to finding any national ‘step back’ in their realisation in violation with Article 2.

Beyond ‘backsliding’

Mader, in particular, stresses that non-regression does not require a de minimis threshold of reduction in the protection of the rule of law, but rather, any reduction should be covered by the principle. This could ‘close an important doctrinal gap’: ⁴³ applied sensitively, the non-regression principle could capture incremental changes that disguise themselves as minor modifications but in reality have a systemic negative impact on the rule of law when compounded with other incremental changes. ⁴⁴

This argument is compelling, especially as it infuses legal doctrine with the insights of political scientists. The idea of ‘non-regression’, already by virtue of synonymity, seems to be fully in step with the ‘democratic/constitutional/rule-of-law backsliding’ paradigm predominant in political science and comparative constitutional law. According to this paradigm, ‘troubled democracies are now more likely to erode rather than to shatter – to decline piece by piece instead of falling to one blow’. ⁴⁵ The ‘backsliding paradigm’ sees democratic erosion as an ‘aggregative process made up of many smaller increments’. ⁴⁶ If the non-regression principle could thus capture such incremental changes before they lead to structural and systemic backsliding, it could seriously contribute to stabilising liberal democracy in the member states.

However, the ‘backsliding’ paradigm comes with weaknesses and overgeneralisations that legal doctrine might equally want to consider. The relevance of the ‘democratic backsliding’ paradigm that arguably informs the idea of non-regression should not be overstretched, as not all political and constitutional changes can reliably be located on a linear trajectory between democracy and autocracy. ⁴⁷ As Licia Cianetti and Seán Hanley argue in a thought-provoking article, developments that might lead some to the conclusion of backsliding might actually reveal ‘non-linear dynamics [that] need to “be understood as alternative directions, not way stations” on a journey between autocracy and democracy’. ⁴⁸ While some developments may be no less worrying – for instance, a form of ‘democratic careening’ between illiberal backlash and democratic bounce-back ⁴⁹ – they might nonetheless force us to reconsider the proper response to these developments from the perspective of EU law. It is difficult to reliably anticipate any legal change as the first increment of a wider, systematic deterioration. Attempting to do so notwithstanding these difficulties might give way to a capricious and haphazard deployment of the ‘regression’ argument.

Especially in countries characterised by political polarisation, some measure of ‘careening’ between liberal and illiberal forces with opposing democratic claims might be nothing short of unavoidable. Such unsettlement does not necessarily ‘[mark] the lead-in to a new and different political game – rather, this struggle between opposing democratic claims is the game’. ⁵⁰ These insights complicate the idea of ‘non-regression’: if there is really no escape from the oftentimes messy competition between opposing liberal and illiberal democratic claims, what is gained from policing regressions, especially if they do not necessarily fall foul of what is minimally required of the member states? If some polities are ‘careening’ rather than ‘backsliding’, might more trust in these countries’ abilities to self-correct be warranted? The example of Slovenia – which had already been slated as the next ‘backslider’ in the EU – shows that backsliding is not unavoidable and member states often have the capacity to regenerate after a period of illiberal governance. ⁵¹

This is not to advocate for blind trust in the resilience of member states’ constitutional systems, let alone minimise those cases where backsliding has clearly occurred. Rather, it is to warn about the weaknesses of an overly sensitive non-regression principle. There might be little sense in trying to arrest an anticipated ‘backsliding’ sequence in a member state by policing ‘regressions’ so long as those regressions do not profoundly draw into question the ability of member states to yield authority to one another.

Non-regression and constitutional experimentation

Worrying constitutional developments can reveal deeper complexities and more difficult trade-offs that cannot be reliably placed on a linear trajectory. Underlying problematic developments can be ‘uncomfortable normative and political choices among stability, inclusivity, and contestation’ ⁵² that render a straightforward diagnosis of ‘regression’ much more complicated. In the same vein, constitutional reforms that might appear concerning might guide a country towards a different trajectory of constitutionalism rather than divorcing it from liberal constitutionalism altogether.

Consider a hypothetical example: suppose a new, ostensibly ‘populist’, political movement sweeps to power in Poland, seeking to reform constitutionalism in the country while drawing lessons from the ‘rule of law crisis’. They look, with particular worry, at the Polish Constitutional Tribunal’s abortion ruling ⁵³ and conclude that the ease with which the judiciary could be captured even from within a liberal constitution meant that the powers of the judiciary should be counterbalanced. Rather than allowing for political questions to be constitutionalised by means of judicial capture, they ought to remain solidly within the political domain, where political institutions allow for a balance of power and agonistic competition between competing political forces. They argue that, in such a politically polarised society, granting the final say to ostensibly ‘neutral’ institutions like the Constitutional Tribunal merely invites their capture and instrumentalisation as a means of political domination. Accordingly, they decide to abolish ‘hard’ judicial review of legislation and instead transition to a more political model of constitutionality control subject to legislative override, potentially modelled upon the ‘notwithstanding clause’ in the Canadian Charter of Rights and Freedoms. ⁵⁴

Critics of such a proposal might not share this assessment of the rule of law crisis in Poland. Not without merit, they might argue that underlying the abuses of the Constitutional Tribunal was quite simply the illegal appointment of ‘quasi-judges’ in cahoots with the government breaking the rules. Furthermore, weakening judicial review powers can easily be considered a ‘regression’: after all, a previously existing strong legal safeguard against unconstitutional legislation has been significantly weakened. Similarly, abolishing the ‘hard’ judicial review of legislation is a clear departure from the model recommended by authoritative advisory bodies such as the Venice Commission. ⁵⁵

But based on the mere weakening of judicial review powers alone, it is hard to tell whether such a change would necessarily lead to a significant deterioration of constitutionalism. The motivation behind the hypothetical reform is not malicious; rather, it comes from a place of sincere concern for stable constitutionalism. Simply treating any instantiation of democratic constraints as ‘progress’ and any removal of such constraints as ‘regression’ results in a dangerously lopsided view of constitutionalism that fails to recognise that different circumstances can warrant different balances to be struck between the rule of law and democracy. While the dangers of weakening judicial review powers are well known, such a move could also (ironically) be beneficial for judicial independence, generating fewer incentives to illegally capture the Constitutional Tribunal. It might encourage political actors to channel their fundamental disagreements through the political institutions designed for them rather than attempt to circumvent them through judicial capture. Should such a constitutional reform raise the alarm bells of ‘regression’, or could it be plausibly defended as a new stage in the evolution of Polish constitutionalism? ⁵⁶

This is not to downplay the dangers of democratic backsliding in times of ‘constitutional acceleration’, ⁵⁷ let alone deny that Poland and Hungary have clearly experienced sustained backsliding at the hands of their respective authoritarian governments. Rather, we need to make sure that our background assumptions around ‘backsliding’ are sufficiently complex and nuanced to inform doctrinal responses to constitutional and legislative changes in the member states. The danger becomes one of overidentifying legislative and constitutional changes as foreshadowing ‘backsliding’. Too great a fear of ‘backsliding’ might counterproductively choke the potential for constitutional innovation, experimentation, and regeneration in the member states.

Value complexities

The previous example illustrates some of the profound complexities one encounters when considering how concretely ‘regressions’ can be identified. The concept of ‘regression’, after all, is exceedingly vague. It does not tell us anything about what specifically is required of the member states in regard to Article 2 values – it merely tells us that there cannot be any less of it. How do we know whether or not a given measure constitutes a ‘regression’ in the protection of Article 2 TEU values?

In one sense, of course, the notion of constitutional regression can be grasped in a visceral way, as those who have followed the dismantlement of checks and balances in Poland and Hungary can tell all too clearly. But such low-hanging fruit should not induce us to overlook the profound complexities that can arise in different situations. Beyond flagrant examples of constitutional backsliding, ‘measuring’ examples of regression that do not necessarily fall foul of minimum standards is an incredibly difficult exercise. The simplicity of the very idea of non-regression should not lead us to gloss over the deep complexities that come with assessing constitutional changes and developments, let alone ‘measuring’ their impact on a simple scale of increasing or decreasing protection for Article 2 values. ⁵⁸

Let us take the rule of law, as the most fleshed out Article 2 value, as an example. Disregarding for a second the ‘essentially contested’ conceptual nature of the rule of law, ⁵⁹ we can accept that a number of essential elements characterise the rule of law as a legal principle in the EU. The European Commission’s working definition of the rule of law thus defines it as a principle according to which ‘all public powers always act within the constraints set out by law, in accordance with the values of democracy and fundamental rights, and under the control of independent and impartial courts’, ⁶⁰ providing a long list of principles such as legality and legal certainty that form part of the rule of law. Such a definition can, then, be further disaggregated into concrete indicators and benchmarks to be found on a wider ‘checklist’. ⁶¹ But this disaggregation does not yield a coherent picture of the rule of law in and of itself. ⁶² Different elements of the rule of law are not always symbiotic – at times, one will come at the expense of another. At times, they will interact with each other in different ways. As Tom Ginsburg argues,

there is no single ideal formula to achieve [the rule of law]. It may be that, in some countries, an independent judiciary is a crucial element; in other countries the judiciary can become too autonomous and can itself become a major political actor. In some countries, prosecutors will be key actors for ensuring that the rule of law is upheld; in others, civil society might be more important. ⁶³

The example of judicial councils is illustrative of this complexity. Self-governing judicial councils were established across many states in Central and Eastern Europe as the ‘best practice’ institutional template for the organisation of the judiciary. ⁶⁴ These councils are often almost entirely insulated from political influence and largely run by the judicial branch itself, granting the highest possible protection of judicial independence. Such high degrees of independence have unforeseen consequences. As Bobek and Kosař have observed, the introduction of self-governing judicial councils in many Central and Eastern European countries without accompanying or preceding ‘cultural change and personal renewal’ has meant that threats to judicial independence no longer came primarily from political actors but instead from within the judicial hierarchy. ⁶⁵ In fact, the judiciary’s insulation from any form of political accountability has encouraged and facilitated forms of politicisation, corporatism, rent-seeking, and ultimately, corruption, within the judicial branch. ⁶⁶

‘More judicial independence’, then, does not necessarily translate into ‘more rule of law’. ⁶⁷ In fact, increased institutional independence of the judiciary can come at the expense of the independence of individual judges. ⁶⁸ Conversely, ‘less of any particular indicator of the rule of law’ ultimately does not always translate into ‘less rule of law’. Measures that might actually be conducive towards the rule of law may look formally regressive on paper. Changing the composition of a judicial council to include politicians as well as judges, for instance, might be regarded as formally reducing judicial independence, but it might also increase the accountability of the judiciary. ⁶⁹

Other examples illustrating the complexity of regression assessments abound. For instance, various proposals that purport to defend or restore the rule of law and democracy may also be seen to violate the rule of law in other ways. The proposal of the Hungarian opposition ahead of the 2022 election to rewrite the Constitution even without an enabling supermajority has similarly been slated as both a major warning sign for regression ⁷⁰ and the only plausible way to restore and revive democracy in Hungary. ⁷¹ Looking beyond Europe, proposals to ‘pack’ the Supreme Court of the United States with a view to democratic restoration rather than capture illustrate the paradoxes that complicate the idea of ‘regression’ in a similar way. ⁷²

Underlying these complexities and trade-offs is the fundamental ambiguity of constitutional reforms. As Mark Tushnet puts it, ‘almost every specific constitutional development can be a valuable reform in some contexts and something that weakens constitutional democracy in other contexts’. ⁷³ Whether or not any of these measures will be ‘regressive’ depends on a vast number of factors beyond the reform itself: the interrelations between the measure in question and the legal and political culture of the polity that undertakes the measure, the concrete political context in which the measure is taken, and, most crucially, the intentions of those who undertake it. Inevitably, the judges’ ideological assumptions and intuitions about the proper relationship between politics and law will similarly shape whether or not a certain measure will be considered ‘regressive’.

All of these factors complicate the idea that constitutional and institutional changes can be mapped on a scale of progress, allowing for juridical findings of ‘regression’. Simplistic narratives of progress and regression eschew the complex and contingent nature of constitutional practice and legitimacy. Non-regression might be too blunt a tool for a task of such considerable complexity as safeguarding the rule of law, let alone other values protected by Article 2 TEU.

Asymmetrical enforcement and the equality of member states

The non-regression principle makes accession to the EU the central anchor point for the enforcement of Article 2 values. Drawing on Wightman, ⁷⁴ the Court argues that, since Article 49 TEU requires member states to be committed to the values in Article 2 TEU, their continued compliance with these values is a ‘condition for the enjoyment of all of the rights’ deriving from EU membership. As the Court puts it in a later case, ‘[c]ompliance with those values cannot be reduced to an obligation which a candidate State must meet in order to accede […] and which it may disregard after its accession’. ⁷⁵ This link, while conceptually sound, is politically perilous: overemphasising the commitments member states made at accession runs the risk of perpetuating the power asymmetries that came with the accession process itself and can therefore potentially threaten the equality of member states.

Non-regression and the shadow of accession conditionality

To understand why this is the case, a closer look at the accession process itself is necessary. Accession to the EU has changed significantly over the years. The Copenhagen Criteria for accession were only laid down by the Council in 1993, and ‘no comparable mechanism operated in the three previous waves of accession’. ⁷⁶ As Sadurski points out, the fifth wave of enlargement carried with it a problem of ‘double standards’: ‘the EU expected, through its political conditionality, more from the applicant states than from its own member states’. ⁷⁷ The member states acceding to the EU after 1993 were held to standards which the founding states and the states acceding before were not held to. Furthermore, accession was characterised by a ‘clear asymmetry in power relations not just between the EU and the CEE candidate countries but also between the old MSs and the aspiring candidates’. ⁷⁸ Linking non-regression to accession not only raises questions concerning the soundness of holding member states to account on the basis of at times questionable pressures and conditionalities that led the then-candidate states to adopt certain institutional arrangements. More crucially, it raises questions about the equality of member states – if fleshed out the wrong way, it could be seen to unduly reproduce the power asymmetry between member and candidate states into asymmetric levels of constitutional autonomy granted to ‘pre-Copenhagen’ and ‘post-Copenhagen’ member states, respectively.

The court’s methodology in Repubblika itself is not very promising in this respect. The explicit link between the non-regression principle and Article 49 TEU in addition to the commitment made to EU values in acceding to the Union suggests that the status quo at the time of accession will play a special role in the way the Court will apply this principle. In fact, the Court goes out of its way to explicitly establish that the provision of the Maltese Constitution in question in the case remained unchanged from 1964 to 2016 and that ‘it was therefore on the basis of the provisions of the Constitution in force prior to that reform that the Republic of Malta acceded to the European Union under Article 49’. ⁷⁹ Despite some ambiguity, ⁸⁰ one cannot help but think that the Court explicitly went back to the point of accession in order to establish that point as the baseline from which an assessment of ‘regression’ should proceed. Otherwise, it could have simply established what the previous constitutional provision stated and proceeded in its regression assessment on that basis.

Of course, within the logic of non-regression, looking to the constitutional provision in force at the time of accession can be considered sensible. After all, some measure of ‘regression’ might have already occurred since the time of accession – little would be won, for instance, by guarding the protection of judicial independence in Poland against regression from the current legislative status quo. Tying regression to the status quo at accession, rather than the immediately preceding legislation, might allow the Court to avoid having to apply a lower bar than possible or necessary.

However, one question immediately arises: what standard are the founding member states held to, who did not undergo an exacting conditionality process to join the Union? What of those member states that joined before the Copenhagen criteria were formulated? ⁸¹ Are their constitutional arrangements’ conformity with Article 2 values taken for granted? ⁸² Taking the point of accession as the baseline from which a regression assessment proceeds would make obvious the glaring asymmetries between ‘pre-Copenhagen’ member states, whose constitutional arrangements got to freely evolve into what they are now, and ‘post-Copenhagen’ member states whose constitutions were subject to benchmarks, indicators, and conditionalities they did not choose themselves.

The alternative, then, would be to simply look to the laws as they were in force before any disputed legislative change was enacted. This would do away with the most glaring inequalities. Nonetheless, the non-regression principle is still bound to lead to an asymmetrical enforcement of Article 2 TEU. Reforms in the member states might be at risk of violating Article 2 TEU because they constitute a ‘regression’ from a previous level of protection of Article 2 values, even though other member states have always had similar constitutional arrangements.

This unequal and asymmetrical enforcement of Article 2 requirements will continue to raise questions. Crucially, these asymmetries will in no small part continue to be shaped by the pressures of accession conditionality. Many of the institutional changes enacted in the post-2004 member states under the pressures of accession conditionality are, after all, still in place. Accession conditionality pushed many Central European states to fortify legalistic constraints and adopt pre-fabricated institutional templates – such as autonomous judicial councils – that often did more harm than good by encouraging judicial corporatism and rent-seeking. ⁸³ In addition, the Commission (even if tacitly) pushed Central and Eastern European states towards a ‘consensus’ model of democracy, regardless of whether salient social and political contingencies were well suited to such a model. ⁸⁴ They have, at least partly under the influence of the EU, adopted strong models of constitutionality control featuring powerful constitutional courts. ⁸⁵ Romania, for instance, was encouraged by the European Commission to abolish a clause allowing a two-thirds majority in Parliament to override constitutional court judgments, since this posed ‘a major obstacle to genuine constitutional control in Romania’. ⁸⁶ However, other member states not subject to conditionality have opted for more political control, both of the judiciary and of the constitutionality of legislation. For instance, the Dutch Constitution explicitly bans the judicial review of legislation, ⁸⁷ while Finland predominantly employs a political model of constitutionality control with a very limited role for decentralised judicial constitutionality control. ⁸⁸ In other words, the post-Copenhagen (‘new’) member states were pushed along a trajectory of ‘constrained democracy’ with empowered judiciaries and weakened legislative power – further than many of the ‘old’ (pre-Copenhagen) member states themselves are comfortable with. Inevitably, this raises questions about where regression begins and whether different member states can be legitimately held to different standards based on their respective starting points.

Referring back to previously discussed examples: would the decision of a member state that joined with firm judicial review of legislation in place to shift to a more political form of constitutionality control be considered a ‘regression’ in the field of the rule of law, despite similar arrangements being acceptable elsewhere? Would the decision of a member state that joined with a self-governing judicial council to allow for a stronger political component in the judicial appointment process be considered a regression, despite other member states having similarly political appointment processes? After all, the Court has not stipulated any de minimis rule for regression – rather, ‘any’ regression of a member state’s laws seems to be at issue. At least by the rule of law standards established by the Venice Commission, moving away from either of these models would entail a departure from the recommended model, ostensibly reducing the protection of the rule of law. ⁸⁹

Asymmetric enforcement of Article 2 requirements – granting a wider space for constitutional autonomy to some states than to others, depending on how extensive their protection of the rule of law already is – is bound to garner scepticism and backlash. The mere fact that a certain level of protection for the rule of law is less than a previous level of protection attained in a member state cannot in and of itself be sufficient to trigger a violation of Article 2 TEU, especially if other member states suffice the same requirements with a similar level of protection. ⁹⁰ Anything else would lead the Court to apply inconsistent double standards at odds with the equality of member states that cannot but endanger the legitimacy of the Court’s value jurisprudence.

In defence of asymmetry?

Some might argue that the looming spectre of a ‘Frankenstate’ that pastes together elements from constitutions in other member states to arrive at an ultimately ‘monstrous’ creation ⁹¹ should give the Court licence to strike down institutional arrangements in one country that work well in another country. Indeed, arguments that the Court or Commission are applying ‘double standards’ often frivolously point to the existence of similar institutional arrangements in wildly different contexts. ⁹² Non-regression, in this context, has been praised as a useful way to debunk ‘Frankenstate’ arguments: what is the status quo in one country might indeed constitute a regressive measure elsewhere and could thereby be declared incompatible with Article 2 TEU. ⁹³ But the Court already has a way of dealing with this type of ‘abusive constitutional borrowing’ ⁹⁴ – it can do so precisely by spotlighting how certain institutional arrangements fall short when compounded by other changes within a particular context. ⁹⁵ It does not need to seek recourse in non-regression to make this point. In fact, the Court’s arguments will only become more convincing the more concretely it engages with contextual differences rather than evading them with reference to ‘non-regression’. Like all arguments from abusive constitutional borrowing, the ‘Frankenstate’ argument should not be falsely turned on its head: suspicion of ‘abusive constitutionalism’ ought not become a blanket argument against institutional solutions that move into a more political, rather than a more legal, direction. ⁹⁶

Secondly, some might argue that asymmetric requirements can be justified on the basis that ‘new democracies’ do not have the requisite constitutional culture and traditions to sustain more ‘political’ institutional arrangements, as the Venice Commission also frequently stresses. ⁹⁷ Such arguments need to be treated with even more caution and scepticism. Systematic recourse to this argument deprives it of the nuance and contextualisation necessary to sustain it, ⁹⁸ rendering it a vague rule of thumb at best, and a tired transitologist trope at worst.

As the example of self-governing judicial councils illustrates, it is far from clear that an alleged ‘lack of constitutional culture’ can be remedied through more rigid and legalistic constitutional arrangements. But even where such arguments could be plausibly advanced, they are ill-suited to the concrete political context of the European Court of Justice interfering in domestic constitutional affairs: they are bound to become a lightning rod for criticism and backlash. They infantilise the (not-so) ‘new’ member states, presenting them as requiring firmer constitutional ‘hand-holding’ for their purported lack of established constitutional culture and traditions, even 30 years after the fall of the Iron Curtain. ⁹⁹ Indeed, it is hardly surprising that the Polish government jumped at the European Court of Human Rights’ mere reference to the ‘new democracies’ argument in Advance Pharma ¹⁰⁰ as a way of discrediting said judgment. ¹⁰¹

‘Minimalist’ non-regression?

The idea of a fully-fledged non-regression principle is marred by a number of problems: the difficulties of identifying ‘regressions’ on the sole basis of legislative provisions, the complex nature of the values for which ‘regression’ is to be identified, the limits of the ‘backsliding’ paradigm that seemingly informs the idea of non-regression for Article 2 TEU, as well as the potential dangers to the equality of member states that come with the link between non-regression and accession. But what about less invasive conceptions of non-regression?

There are, after all, good reasons to think that the Court did not intend to dramatically deepen the intensity of review by introducing a non-regression principle – maximalist visions of non-regression do not sit well with the existing case law of the European Court of Justice. For instance, in the Disciplinary Regime case, the Court considered the changes undertaken by the Polish government in a wider context rather than looking at them incrementally. It took explicit note of the wider reforms of the Polish judiciary and assessed whether their compounded effect produced a situation contrary to EU law. Only ‘in the context of an overall analysis’, ¹⁰² rather than on the basis of any single change evaluated in isolation, did the Court find that the requirements stemming from Articles 2 and 19 TEU were not met. This suggests that the threshold of sensitivity for violations is significantly higher than a maximalist version of non-regression would suggest.

A higher threshold for regression?

Other commentators have indeed conceived of non-regression in such less invasive ways. Spieker, for instance, imagines that the non-regression principle should not capture minor changes to the member states’ institutional configurations, but only ‘significant regression[s] from pre-existing national standards’. ¹⁰³ Such an approach seems more closely aligned with existing case law. A higher threshold of sensitivity for regression findings could also help the Court navigate some of the pitfalls raised in the first half of this article, but it still requires an answer to the problem of asymmetrical enforcement. If the baseline from which ‘significant regressions’ are assessed is defined by the current level of protection for the rule of law provided by a member state, the question of double standards still looms large. A change that might constitute a significant regression in one member state might still reflect the status quo in another member state. It is not clear how the Court of Justice can defend upholding such asymmetrical enforcement of Article 2.

Increasing the threshold of severity also raises questions of definition and demarcation: what differentiates a significant regression from a non-significant one? Is it merely a certain amount of deterioration, regardless of the resulting level of protection for the rule of law? And if so, how can the Court ensure consistency in its findings to this effect? The vagueness of the non-regression test, Spieker argues, is a positive, allowing the Court to avoid providing a ‘full-blown account’ of Article 2 TEU, and thereby helping to preserve constitutional diversity among the member states. ¹⁰⁴ But too much vagueness can also play against the Court’s favour. While Article 2 should be developed in a sufficiently open-textured manner to avoid prescribing concrete institutional solutions to the member states, excessive vagueness might deprive Article 2 requirements of the substance required for a sufficiently consistent doctrine. A test for conformity with requirements stemming from Article 2 TEU should not be so substantively thin that it could be interpreted as a carte blanche for the Court.

An anchor for minimum requirements?

Alternatively, then, one might ask whether there might be certain hallmarks of significant regression regardless of the pre-existing level of protection. In fact, this might be what the Court had in mind when it stated in Repubblika that member states must not adopt rules ‘which would undermine the independence of the judiciary’ ¹⁰⁵ – a binary test rather than a sliding scale. The ‘non-regression principle’ might, then, simply be the basis on which the Court establishes more concrete requirements that the member states must fulfil. This seems to be a more reasonable way of making sense of non-regression: the Court would oblige itself to provide some form of yardstick as to what makes a regression significant. It would also alleviate concerns about the equality of member states, as the Court would – at least formally – apply the same yardstick to all, regardless of a member state’s previous level of protection.

However, such a modification would make the ‘non-regression principle’ a misnomer: the Court would no longer ask whether the new law is worse than the old one, but whether the new law is so bad that judicial independence is undermined. ‘Non-regression’ would become a moot point, effectively replaced with binary prohibitions or abstract minimum requirements. Where minimum requirements exist, the regression exercise the Court undertook in Repubblika is simply not necessary: if all that matters is whether a constitutional or legislative arrangement crosses the ‘red lines’ posed by Article 2 TEU, there is no need to compare two sets of rules. It simply needs to be established whether or not the red line has been crossed. All that would be left of non-regression is the necessity to comply with the values in Article 2 on the basis of the ‘voluntary commitment’ the member states have made.

What place for non-regression?

Many questions remain about the position that a non-regression principle for Article 2 might take within the existing case law, and which of these options (a full-blown non-regression principle, whether limited to ‘significant’ regressions or not, or non-regression as an anchor for minimum standards) the Court will choose remains to be seen. In particular, how tests for minimal requirements (like the ‘appearances’ test) are to be reconciled with non-regression remains unclear. The danger is that one will constantly displace the other. Either a regression will only be found where a member state has amended its laws in a way that will affect its compliance with minimum requirements – in which case ‘regression’ is a moot point – or a finding of regression in the laws of a member state will constitute a violation of EU law regardless of any posited minimum requirements.

Two cases in which this test has been applied – Repubblika itself, and Commission v Poland (Régime disciplinaire des juges) illustrate this dilemma, and show that perhaps the Court itself does not quite know what to do with the idea it conjured. In Repubblika, the Court took no issue with the reforms in question, as they were, firstly, found to ‘reinforce the guarantee of judicial independence’ ¹⁰⁶ (and were thus not ‘regressive’), and, secondly, passed muster under the ‘appearances test’. ¹⁰⁷ Thus, the Court first ascertained whether the new laws reduced the protection of the rule of law in Malta compared to the rules in force at the time of accession, and then made sure that no minimum requirements had been violated. Had the Court found a ‘regression’, one might assume that the question of conformity with minimum standards would have been rendered immaterial.

In Commission v Poland, ¹⁰⁸ the Court similarly applied both tests but switched the order in which they were referred to. The Court first established that ‘the important role played by the KRS […] in appointing members of the disciplinary chamber […] gives reasonable doubts in the minds of individuals as to the independence and impartiality of that chamber’, ¹⁰⁹ and only later adds the statement that ‘such a development constitutes a reduction in the protection of the value of the rule of law’. ¹¹⁰ Why was a finding of regression even necessary here? The Court had already established that the minimum requirements for judicial independence had not been met. If ‘regression’ is just a synonym for ‘not meeting the minimum requirements’, it becomes substantively insignificant – a duplication of more substantive tests that either explicitly exist alongside non-regression, or a vessel for requirements that are not explicitly spelled out by the Court. How exactly the Court will reconcile non-regression with more concrete and substantive requirements thus remains to be seen.

It may be the case that the Court wants to retain both options – develop minimum requirements in fields such as judicial independence and, where the latter proves difficult or controversial, make more contextual findings based on non-regression. This, however, might prove a difficult line to legitimately maintain. The Court would need to explain why, in some situations, it resorts to explicit substantive standards while leaning on the substantively empty non-regression test in other situations. This would become especially problematic where the Court opts for non-regression instead of the substantive requirements the Court already established: if the Court cannot convincingly apply the substantive requirements that it previously formulated, resorting to non-regression to identify a violation (or no violation) of Article 2 TEU would appear highly questionable.

Conclusion

The non-regression principle formulated by the European Court of Justice in Repubblika has sparked the enthusiasm and imagination of many legal scholars seeking to judicially flesh out the values of Article 2 TEU – Leloup, Kochenov, and Dimitrovs even called it a ‘grand opening, marking something new and potentially truly far-reaching’. ¹¹¹

It was the intention of this piece to temper that enthusiasm and encourage more critical reflection about the many problems and pitfalls that would come with a veritable non-regression principle for Article 2 values. While ‘non-regression’ promises to ‘solve’ the Copenhagen dilemma, allowing the EU to continue to shape constitutional developments in the member states and thus prevent them from turning their back on liberal democracy, this article suggests that the ‘end of history’ is not as easily constitutionalised. Non-regression is bound to be confronted with the flaws that come with the ‘progress trajectory’ implicit in it, the complexities of assessing measures as ‘regressive’, and its potentially problematic ramifications for the equality of member states. In that vein, this article has identified at least three major pitfalls that the non-regression principle faces.

The first pitfall pertains to the limits of the backsliding paradigm that arguably informs the idea of non-regression. It is less concerned with the substance of the non-regression principle than with the political imagination behind that principle, which is bound to influence the way in which it is invoked. The ‘non-regression’ principle is embedded in predominant narratives of ‘constitutional backsliding’, according to which minor, incremental changes can gradually erode constitutional democracy altogether. This ‘backsliding paradigm’ certainly has its place – but its validity is limited by more complex and nuanced developments that eschew a linear trajectory. Once we recognise its limits, reliance on ‘non-regression’ becomes more problematic: trying to identify the first ‘regressive’ increment of a future sequence of backsliding could give way to capricious enforcement. The European Court of Justice could expend normative resources in halting minor changes capable of being remedied from within a member state’s constitutional system. More than that, too much fear of ‘backsliding’ might run the risk of impoverishing our potential for constitutional experimentation, renewal, and regeneration through ambitious constitutional reforms at the national level. ¹¹²

The second pitfall is closely related and pertains to the value complexities that complicate simple findings of non-regression. Not all ‘steps back’ pertaining to safeguards for certain values necessarily endanger those values. The underlying problem is the fundamental ambiguity of constitutional reforms that arises when inspecting them with respect solely to their institutional or formal aspects. Given the right intentions, even measures that formally look like they would reduce the protection of the rule of law, for instance by diluting protections of judicial independence, might actually be conducive to the rule of law – the institutional deficiencies of self-governing judicial councils provide a case in point here. These complexities might render simple ‘regression’ assessments a lot more complicated.

The final, and perhaps most important, pitfall pertains to the asymmetrical enforcement inherent in the non-regression principle and the resulting threats to the equality of member states in the EU. The differential treatment of member states is intrinsic to the non-regression principle: using the constitutional and legislative situations in the respective member states as a baseline, inevitably some things will be struck down as ‘regressive’ in some member states that could be considered acceptable in others. What may be intended as careful deference to local standards could turn into a source of great political peril for the Court: the fault lines of ‘regression’ are bound to be shaped by the pressures of accession conditionality imposed on the member states that joined after the Copenhagen criteria were laid down. The result could be a significantly slimmer space of constitutional autonomy for the ‘post-Copenhagen’ member states than for those that joined earlier – a discrepancy that seems at odds with the equality of member states.

This close alignment of the non-regression principle with Article 49 TEU, the accession process, and the conditionalities that came with it, is one of its most considerable weaknesses. Rather than emancipating the enforcement of Article 2 values from the power asymmetries of accession – which have at times given credence to accusations of ‘double standards’ and even ‘neo-imperialism’ ¹¹³ – non-regression might bring the enforcement of EU values closer to these dangers to its legitimacy. ¹¹⁴ Unless the same standard applies to all member states, value enforcement will run the risk of entrenching the inequalities that were supposed to have evaporated once candidate states became member states.

All of this leaves more questions than answers in regard to non-regression. If the Court is to proceed with non-regression as a guiding principle for the enforcement of Article 2 TEU, it needs to find ways of navigating these pitfalls. However, the more these pitfalls are circumvented, the less is left of the idea of non-regression. All that might remain of the ‘non-regression principle’ is the simple link between commitment and obligation under Article 2 TEU.

One might object that some of these pitfalls could be applied to the judicial enforcement of Article 2 TEU altogether, rather than merely non-regression, and that taking them seriously, in fact, draws all of EU value enforcement into question. However, the pitfalls discussed here resonate in particularly troublesome ways with regard to the non-regression principle and the problematic notions of ‘constitutional progress’ and ‘regression’ it evinces. This does not ‘burn down the house’ of EU value enforcement, as there is a fundamental difference between policing regressions and enforcing minimum requirements: the latter makes sure that national constitutional orders meet the baseline level of legitimacy required to sustain the authority of EU legal order, while the former potentially does something quite different altogether.

Nonetheless, the pitfalls raised here bring up complex and, at times, uncomfortable issues that those who seek to judicially flesh out Article 2 into different directions need to consider. How should the contingency and context-sensitivity of different constitutional arrangements influence the way in which the European Court of Justice passes (or should refuse to pass) judgment about them? How does our imagination of Article 2 values change once we think beyond the ‘backsliding’ cases of Poland and Hungary? Does our fear of backsliding excessively constrain our ability to imagine constitutional alternatives and new ways of making sense of the values in Article 2? Does the path to a union of values mean doubling down on old orthodoxies regarding those values, or does it require new thinking from the ground up?

There is little doubt in the author’s mind that the European Court of Justice has a role to play in safeguarding the legitimacy of the EU as a polity in the light of severe transgressions of the values protected by Article 2 TEU. This is particularly so where such transgressions undermine mutual trust, eroding the ability of member states to yield authority to one another and, by extension, to the EU. ¹¹⁵ How far that role should extend, and whether non-regression is the right way to give expression to that role, however, is another question altogether.