‘Integration-through-Law’: grand theory, revisionist history

Abstract

How has the European Union been integrated in the past? Legal academics have traditionally pointed to the Court of Justice and to the broader idea of an ‘integration-through-law’. Through its supranational jurisprudence, the Court – not the EU legislature – was thus placed at the centre of the European integration project. The underlying reasons for this dominance of constitutional ‘law’ over legislative ‘politics’ have thereby been the subject of three famous explanations: the ‘equilibrium theory’ (Weiler), the ‘asymmetry theory’ (Scharpf) and the ‘over-constitutionalisation theory’ (Grimm). What are the merits of these grand theories of European integration when measured against the historical record? This article hopes to explore this question in the context of the internal market. Its historical revision begins with an analysis of the respective spheres of normative and decisional supranationalism during and after a foundational period (Sections 2 and 3). This is followed by an examination of the meaning and significance of the Cassis de Dijon judgment in the late 1970s. Through this revolutionary case, a dialectical relationship between the EU Court (‘law’) and the EU legislator (‘politics’) emerges (Section 4) that ultimately leads to the spectacular rise of EU legislation (Section 5) after the SEA. This transformational relationship will provide the critical lens for a historical revaluation of the three grand theories of legal integration (Section 6).

1. Introduction

How has the European Union been integrated in the past; and what are the respective roles of ‘law’ and ‘politics’ in this process? The traditional answer has here insisted on a constitutional dominance of the judicial over the political sphere, with especially legal academics having overwhelmingly pointed to the Court of Justice and the idea of an ‘integration-through-law’. ¹ Through its supranational jurisprudence, the EU Court – not the EU legislature – is here seen to be at the centre of the earlier integration project.

The reasons behind this apparent triumph of judge-made ‘law’ over legislative ‘politics’ have been explored by three famous integration-through-law theories. In 1981, a young Joseph Weiler proposed, for the first time, his ‘equilibrium theory’. ² According to that theory, normative supranationalism (‘law’) and decisional supranationalism (‘politics’) were locked in an inverse relationship that guaranteed the stability and legitimacy of the ‘Community system’. In essence: the more the normative quality of European law had assumed supranational characteristics via direct effect and primacy, the less supranational the political decision-making process had become. ³ For when the revolutionary judgments of Van Gend and Costa transformed ‘soft’ (international) into ‘hard’ (supranational) law, the Court removed a (selective) ‘exit’ option from the Member States and the latter had, therefore, naturally reclaimed their ‘voice’ in the legislative process. ⁴ This dialectical counter-movement meant that the ‘Community system’ had returned to an intergovernmental-decisional arrangement close to the United Nations, while its supranational-normative system approached that of the United States. ⁵

Taking Weiler’s inverse correlation between supranational law and intergovernmental politics as a starting point, Fritz Scharpf subsequently developed – a decade later – his ‘asymmetry theory’. ⁶ According to him, European integration was profoundly imbalanced as regards the spheres of negative integration (‘law’) and positive integration (‘politics’). He famously explains:

The main beneficiary of supranational European law has been negative integration. Its basic rules were already contained in the ‘primary law’ of the Treaties of Rome. From this foundation, liberalization could be extended, without much political attention, through interventions of the European Commission against infringements of Treaty obligations and through the decisions and preliminary rulings of the European Court of Justice. By contrast, positive integration depends upon the agreement of national governments in the Council of Ministers; it is subject to all of the impediments facing European intergovernmental policy making. This fundamental institutional difference is sufficient to explain the frequently deplored asymmetry between negative and positive integration in EC policy making. The most likely result is a competency gap, in which national policy is severely restrained in its problem-solving capacity, while European policy is constrained by the lack of intergovernmental agreement. ⁷

Importantly, Scharpf’s theoretical claim here partly differs from Weiler’s equilibrium theory. For while both identify an asymmetric development between ‘law’ and ‘politics’ within the Union system, Scharpf sees the Court as an agent of decisional supranationalism; and it is from within that different perspective that three constitutional asymmetries are identified. There is – first – a decisional asymmetry between the supranational organ of negative integration (the Court) and the intergovernmental organ of positive integration (the Council). There is, secondly, a normative asymmetry between negative and positive integration due to the ‘impossibility of political correction’ by the Union legislator once the Court has offered an interpretation of EU law. ⁸ Finally, there is also a competence asymmetry according to which the scope of negative integration is wider than the scope of positive integration.

The success of both integration-through-law theories has been remarkable. They dominate Europe’s legal imagination up to today. ⁹ Their overwhelming influence has even given rise to a third variant in the last decade: Dieter Grimm’s ‘over-constitutionalisation theory’. ¹⁰ The latter directly builds on the ideas of both Weiler and Scharpf by claiming that through the rise of normative supranationalism ‘the direct participation of Member States was no longer needed in order to establish the single market’, because ‘[d]irect effect and supremacy of European law allowed . . . the ECJ (the organ charged with determining the meaning of the treaties in concrete cases) to take the task of implementing economic integration into [its] own hands’. ¹¹ Yet Grimm here also adds a new theoretical claim that goes beyond those of Weiler and Scharpf:

Different from national constitutions, the treaties are not confined to those provisions that reflect the functions of a constitution. They are full of provisions that would be ordinary law in the Member States. This is why they are so voluminous. As long as the treaties were treated as international law, this was not a problem. As soon as they were constitutionalised, their volume became problematic: in the EU the crucial difference between the rules for political decisions and the decisions themselves is to a large extent levelled. The EU is over-constitutionalised. This has two important consequences. First, the over-constitutionalisation severely limits the Member States’ role as ‘Masters of the Treaties’…Second, combined with the lack of differentiation between the constitutional law level and the ordinary law level, the constitutionalisation of the treaties immunises the Commission and particularly the ECJ against any attempt by the democratically responsible institutions of the EU to react to the Court’s jurisprudence by changing the law. ¹²

What are we to make of these three grand theories of European legal integration? This article wishes to critically re-evaluate them through a close historical re-construction of the legal integration of the internal market. ¹³ Why the internal market? Because until at least ‘1992’, if not ‘2007’, the common market was at the centre of European integration. Indeed, after the failure of the European Political Community in 1954, it was through ‘the market’ that the founding fathers had hoped to re-launch a federal Europe; ¹⁴ and the very essence of the 1957 European Economic Community (EEC) had therefore been defined in common-market-terms. Against this backdrop, it is hardly surprising that (almost) all classic EU landmark cases were embedded in an internal market context. ¹⁵ Any grand theory of European integration that cannot explain the past evolution of the internal market thus cannot really claim to properly theorise European integration writ large.

This article pursues its historical-revisionist aim in five steps. Section 2 begins with a reconstruction of the original balance between normative and decisional supranationalism, as well as negative and positive integration, in the foundational period between 1958 and 1969. It will be argued that the EEC Treaty contained, from the very beginning, directly effective legal norms; yet that the main method of establishing the common market was via the political organs of the Union. This original division of power between ‘law’ and ‘politics’, however, changed after the transitional period had ended. In this second period, discussed in Section 3, the removal of all discriminatory national laws could, from now on, be done directly through the EEC Treaty, whereas positive integration was to refocus on the removal of obstacles to trade arising from legislative disparities between the Member States.

This post-foundational balance between the spheres of ‘law’ and ‘politics’, however, radically changed in 1979 with the Court’s Cassis judgment. ¹⁶ With it, the Court came to push negative integration (‘law’) into what was originally designed as the province of positive integration (‘politics’). And yet, as Sections 4 and 5 hope to demonstrate, this did not mean that positive integration, and with it the Union’s political and legislative process, would be asymmetrically disadvantaged. On the contrary, the ultimate constitutional result of the Cassis revolution was more – not less – decisional supranationalism and positive integration. For after the Single European Act (SEA), there will be an enormous rise of EU legislation with legislative ‘politics’ even enjoying applicative priority over EU primary ‘law’.

This self-reinforcing relationship between the EU Court (‘law’) and the EU legislator (‘politics’) undermines, however, all those integration theories that insist on a negative asymmetry between ‘law’ and ‘politics’ in the Union legal order. It will thus be argued in Section 6 that Weiler’s equilibrium theory not only fails to account for the original chronology and place of direct effect in the early Union legal order; his allegedly inverse relationship between ‘law’ and ‘politics’ must itself be inversed. The same is, partly, true for Scharpf (and Grimm), whose asymmetry theories will not only be shown to significantly undervalue the decisional and democratic evolution of the EU legislature since 1957, their postulated zero-sum game between ‘law’ and ‘politics’ will also be judged severely reductionist. A Conclusion will briefly reflect on these results and, equally briefly, outline a ‘republican’ legitimation strategy for the Court’s historical integration-through-law approach.

2. The foundational period: building the common market, 1958–1969

The 1957 Rome Treaty’s sole central task had been the creation of a common market; ¹⁷ and to achieve it, the European Economic Community (EEC) pursued a dual strategy: negative and positive integration. ¹⁸ Negative integration here refers to the removal of obstacles to trade by the European Court (‘law’), whereas positive integration means the adoption of secondary law by the European legislature (‘politics’). Through these two strategies, a common market was to be built during a transitional period of 12 years (divided into three distinct stages), ¹⁹ with the expiry of that period set to be ‘the latest date by which all the rules laid down [in the EEC Treaty] must enter into force and all the measures required for establishing the common market must be implemented’. ²⁰ The creation of the common market was thus to be done gradually and, at first, primarily through the political process.

Nevertheless, and so as to prevent legal retrogressions during the transitional period, the Rome Treaty contained a number of ‘standstill’ provisions that were designed to lock in the 1957 status quo. The EEC Treaty articles on the customs union, for example, made a distinction between ‘new’ (post-1957) and ‘old’ (pre-1957) customs duties: the former were prohibited outright by ex-Article 12 EEC, whereas the latter were to ‘be progressively abolished’ by the political process during the transitional period. ²¹ A similar distinction was made in the chapter on quantitative restrictions on goods. Here ex-Article 30 EEC set out the general prohibition for ‘old’ import restrictions, which were to ‘be abolished by the end of the transitional period’; ²² whereas ex-Article 31 EEC immediately prohibited Member States ‘from introducing between themselves any new quantitative restrictions or measures having equivalent effect’. ²³

An automatic prohibition for new national restrictions could equally be found for the freedoms of persons, services and capital. ²⁴ For the freedom of establishment, ex-Article 53 EEC thus stated: ‘Member States shall not introduce any new restrictions on the right of establishment in their territories of nationals of other Member States’; ²⁵ and, as regards services, ex-Article 62 EEC added: ‘Member States shall not introduce any new restrictions on the freedom to provide services which have in fact been attained at the date of the entry into force of this Treaty’. ²⁶ (For the free movement of capital, the EEC Treaty had preferred a softer obligation: ‘[t]he Member States shall endeavour to avoid introducing within the Community any new exchange restrictions on the movement of capital’. ²⁷ )

By contrast, as regards ‘old’ national restrictions, all three fundamental freedoms expressly acknowledged the end of the transitional period as the date by which all these restrictions were to be removed. ²⁸ The 1957 EEC Treaty thus consistently followed a dual logic of trade liberalisation: Treaty-based non-regression clauses (‘law’) were combined with legislative liberalisation competences (‘politics’) to establish the common market until the end of the transitional period in 1970. Let us therefore look at this new-old distinction in more detail.

A. ‘New’ trade restrictions: negative integration and the European Court

What was the Rome Treaty’s legal regime for ‘new’ restrictions to free movement? In its 1961 judgment in Commission v Italy, the Court had clarified, early on, that the Treaty’s standstill provisions would give rise to automatic – that is: direct ²⁹ – obligations under the EEC Treaty. ³⁰ This was confirmed in Commission v Luxembourg & Belgium (Gingerbread) in 1962. ³¹ Dealing with ex-Article 12 EEC specifically, the Court here left no doubt that the introduction of new customs duties after 1957 violated the EEC Treaty directly. ³² With regard to such self-executing prohibitions, the Member States could not invoke the absence of (implementing) Community legislation during the transitional period.

But if the Rome Treaty contained such directly effective prohibitions, could their legal effects also penetrate the national legal orders? This very question was famously asked and answered in Van Gend & Loos. ³³ The Court here again dealt with ex-Article 12 EEC – the standstill clause prohibiting the introduction of new customs duties. The Netherlands appeared to have violated it and Van Gend & Loos – a Dutch trading company – had therefore brought proceedings in a Dutch court to enforce it. Yet could a private party enforce a self-executing norm of an international treaty in a national court? The European Court of Justice indeed confirmed this possibility in 1963:

The objective of the EEC Treaty, which is to establish a common market, the functioning of which is of direct concern to interested parties in the Community, implies that this Treaty is more than an agreement which merely creates mutual obligations between the contracting States. . . . The conclusion to be drawn from this is that the Community constitutes a new legal order of international law for the benefit of which the States have limited their sovereign rights, albeit within limited fields, and the subjects of which comprise not only Member States but also their nationals. Independently of the legislation of Member States, Community law therefore not only imposes obligations on individuals but is also intended to confer upon them rights which become part of their legal heritage. These rights arise not only where they are expressly granted by the Treaty, but also by reason of obligations which the Treaty imposes in a clearly defined way upon individuals as well as upon the Member States and upon the institutions of the Community. ³⁴

Van Gend is justly celebrated as a constitutional landmark; yet the concept of direct effect was – contrary to a common view – not a revolutionary invention that started in 1963. ³⁵ It was, on the contrary, firmly embedded in the normative structure of the Community legal order. For not only had the latter expressly envisaged the direct effect of its secondary law for ‘regulations’, ³⁶ special parts of the EEC Treaty had clearly been designed with direct effect in mind. ³⁷ Yet with Van Gend, the Court had, crucially, clarified that even those Treaty provisions that were addressed to the Member States could be part of those legal norms giving rise to individual rights that national courts were to enforce. And more fundamentally still, the iconic constitutional vision behind the judgment (‘new legal order’) announced an unconditional departure from the ‘old’ legal order of international law. ³⁸

What, however, were the conditions for direct effect, as set out in Van Gend? The Court here focused on the text of ex-Article 12 EEC and found:

The wording of [ex-]Article 12 [EEC] contains a clear and unconditional prohibition which is not a positive but a negative obligation. This obligation, moreover, is not qualified by any reservation on the part of the States, which would make its implementation conditional upon a positive legislative measure enacted under national law. The very nature of this prohibition makes it ideally adapted to produce direct effects in the legal relationship between Member States and their subjects. The implementation of [ex-]Article 12 [EEC] does not require any legislative intervention on the part of the States. The fact that under this Article it is the Member States who are made the subject of the negative obligation does not imply that their nationals cannot benefit from this obligation. ³⁹

The test for direct effect was here clearly presented: wherever the Treaty contained a prohibition that was ‘clear’, ‘unconditional’ and ‘unqualified’ then an article would have direct effect. This seemed to be a very strict test; yet all three conditions were satisfied as regards ex-Article 12 EEC, with the Court even finding the provision to be ‘ideally adapted’ to having direct effect. One year later, in Costa v ENEL the same conclusion was reached for the Treaty’s other standstill clause in the context of the freedom of establishment:

By [ex-]Article 53 [EEC] the Member States undertake not to introduce any new restrictions on the right of establishment in their territories of nationals of other Member States, save as otherwise provided in the Treaty. The obligation thus entered into by the States simply amounts legally to a duty not to act, which is neither subject to any conditions, nor, as regards its execution or effect, to the adoption of any measure either by the States or by the Commission. It is therefore legally complete in itself and is consequently capable of producing direct effects on the relations between Member States and individuals. Such an express prohibition which came into force with the Treaty throughout the Community, and thus became an integral part of the legal system of the Member States, forms part of the law of those States and directly concerns their nationals, in whose favour it has created individual rights which national courts must protect. ⁴⁰

But so much more than that! Indeed, the Costa Court now also famously spelled out that all directly effective European Community law would enjoy primacy over national law:

The obligations undertaken under the Treaty establishing the Community would not be unconditional, but merely contingent, if they could be called in question by subsequent legislative acts of the signatories. . . . The precedence of Community law is confirmed by [ex-]Article 189 [EEC], whereby a regulation ‘shall be binding’ and ‘directly applicable in all Member States’. This provision, which is subject to no reservation, would be quite meaningless if a State could unilaterally nullify its effects by means of a legislative measure which could prevail over Community law. It follows from all these observations that the law stemming from the Treaty, an independent source of law, could not, because of its special and original nature, be overridden by domestic legal provisions, however framed, without being deprived of its character as Community law and without the legal basis of the Community itself being called into question. ⁴¹

The direct effect and primacy of the EEC Treaty’s standstill clauses on new restrictions was subsequently confirmed for all but one of its fundamental freedoms. ⁴² And, nota bene: because the direct effect of these Treaty provisions ultimately depended on them being ‘unconditional’, that is: not requiring further legislative action by the Community (or the Member States), the normative supranationalism of ‘law’ did here in no way affect the original sphere of ‘politics’ in which the Community’s political organs were to adopt positive legislation for ‘old’ trade restrictions – to which we must now turn.

B. ‘Old’ trade restrictions: liberalisation through positive legislation

Following the original Treaty logic, most pre-1957 restrictions had to be removed, during the transitional period, by means of the political process. ⁴³ Community legislation was thereby to be adopted, after an initial stage, by qualified majority voting in the Council; ⁴⁴ and the Rome Treaty had, with one exception, here consistently chosen the ‘directive’ as its legislative instrument of market integration. ⁴⁵ A good illustration of this general design can be found in the context of the freedom of establishment, where ex-Article 54 EEC stated:

1. 1. Before the end of the first stage [of the transition period], the Council shall, acting unanimously on a proposal from the Commission and after consulting the Economic and Social Committee and the [Parliament], draw up a general programme for the abolition of existing restrictions on freedom of establishment within the Community. The Commission shall submit its proposal to the Council during the first two years of the first stage. The programme shall set out the general conditions under which freedom of establishment is to be attained in the case of each type of activity and in particular the stages by which it is to be attained.

2. 2. In order to implement this general programme or, in the absence of such programme, in order to achieve a stage in attaining freedom of establishment as regards a particular activity, the Council shall, on a proposal from the Commission and after consulting the Economic and Social Committee and the [Parliament], issue directives, acting unanimously until the end of the first stage and by a qualified majority thereafter. ⁴⁶

Following ex-Article 54 EEC to the letter, the Community adopted, in 1962, its ‘General Programme for the Abolition of Restrictions on Freedom of Establishment’ in which it set out a concrete timetable for future Community legislation. And by 1964, the Council had actively begun to make use of its competence through – for example – the adoption of a directive that harmonised the conditions under which Member States could justify restrictions to the free movement of persons on grounds of public policy, public security or public health. ⁴⁷ Parallel legislative endeavours were undertaken for the other fundamental freedoms during the first and second stage of the transitional period too. ⁴⁸

Yet just when the Community was about to pass into the third and final phase of this period on 1 January 1966, ⁴⁹ destiny struck. The Commission had made a daringly federalist proposal on the financing of the ‘Common Agricultural Policy’ – a special part of the EEC common market – that had linked the reform of the Community’s own resources system to the rise of the European Parliament. ⁵⁰ Incensed by this supranational move, and objecting to the start of majority voting in general, Gaullist France began to boycott the Community’s political organs until a ‘compromise’ had been found that would counter-balance the (imminent) transition to decisional supranationalism with France’s national interests. That compromise became known as the ‘Luxembourg Compromise’. ⁵¹ After it, decision-making in the Council would take place under the ‘shadow of the veto’; ⁵² and this new constitutional convention would transform Community decision-making into intergovernmental ‘politics’ for almost two decades. ⁵³

But the legislative programme for the establishment of the common market would slowly continue; and it was to even arrive at a significant achievement in 1968 as regards the free movement or workers. For here, despite the Luxembourg Compromise, the Council had managed to adopt two major legislative instruments before the end of the transitional period. ⁵⁴ The core provision within the ‘Freedom of Movement for Workers’ Regulation thereby stated:

1. 1. A worker who is a national of a Member State may not, in the territory of another Member State, be treated differently from national workers by reason of his nationality in respect of any conditions of employment and work, in particular as regards remuneration, dismissal, and should he become unemployed, reinstatement or re-employment.

2. 2. He shall enjoy the same social and tax advantages as national workers…

3. 3. Any clause of a collective or individual agreement or of any other collective regulation concerning eligibility for employment, employment, remuneration and other conditions of work or dismissal shall be null and void in so far as it lays down or authorizes discriminatory conditions in respect of workers who are nationals of the other Member States. ⁵⁵

This well-known provision successfully ‘translated’ ex-Article 48 EEC into a legislative form that was designed to have – vertical and horizontal – direct effects in the national legal orders. ⁵⁶ This legislative translation did, however, not happen for all fundamental freedoms before the end of the transitional period; and the question therefore arose what effects these freedoms would have, with regard to ‘old’ measures, on and after 1 January 1970.

3. After the transitional period: expanding direct effect, 1970–1974

The EEC Treaty’s common market regime distinguished, as we saw in Section 2, between ‘new’ and ‘old’ trade restrictions. The former were to be automatically prohibited by the Treaty (negative integration), whereas the latter were generally meant to be removed by the political process (positive integration) before 1970. ⁵⁷ Yet was that legislative implementation a constitutional prerequisite and condition; or would the direct effect (and primacy) principles, developed in the context of ‘new’ trade restrictions, also extend to ‘old’ trade restrictions once the transitional period had ended?

The text of the Rome Treaty seemed to give a clear answer to this question: ‘the expiry of the transitional period shall constitute the latest date by which all the rules laid down must enter into force and all the measures required for establishing the common market must be implemented’. ⁵⁸ And it was, consequently, widely assumed that the establishment of the common market would ‘not, under any circumstances, depend on the achievement of certain objectives or the fulfilment of the obligations imposed by the Treaty’ because the ‘absolute automatism of [ex-]Article 8’ meant that all (but one) fundamental freedoms were to get direct effect. ⁵⁹ That this was indeed the case was soon confirmed.

A. Negative integration: general direct effect of the fundamental freedoms

The view that ex-Article 8 EEC would trigger the direct effect of the common market freedoms had gained early judicial support in 1966. For in Lütticke, the ECJ had confirmed that the expiry of the (shorter) transitional period for the prohibition on discriminatory national taxation had indeed led to the full direct effect of the relevant Treaty article. ⁶⁰ This judicial confirmation was subsequently extended to customs export charges under ex-Article 16 EEC; ⁶¹ and it equally informed other parts of the young Community legal order. ⁶² Yet would the expiry of the transitional period really lead to the direct effect of all fundamental freedoms, even as regards ‘old’ trade restrictions?

That this was indeed so was clarified as early as 1970 in SACE. ⁶³ The Court here held:

[Ex-]Article 13(2) [EEC] imposes on Member States the obligation to abolish progressively ‘during’ the transitional period charges having an effect equivalent to customs duties on imports. Although the Commission had to decide the rhythm with which such charges had to be abolished during the transitional period, nevertheless it appears from the very wording of [ex-]Article 13 that these duties had in any event to be eliminated at the latest at the end of the said period. Therefore from the end of this period [ex-] Article 9 must have its full effect on its own. [Ex-]Articles 9 and 13(2) taken together, involve, at the latest at the end of the transitional period, with regard to all charges having an effect equivalent to customs duties on imports, a clear and precise prohibition on exacting the said charges, which is not subject to any reservation for the States to subject its implementation to a positive act of national law or to an intervention by the institutions of the Community. It lends itself, by its very nature, to producing direct effects in the legal relations between Member States and their subjects. ⁶⁴

But what about the rest of the fundamental freedoms? In record time, the Court here also endorsed the general and full direct effect of the Treaty freedoms on goods, ⁶⁵ workers, ⁶⁶ establishment, ⁶⁷ and services. ⁶⁸ By ‘1974’, the sole exception only remained the freedom of capital, whose ‘text’ was judged to be too qualified to ever give rise to direct effects. ⁶⁹

What was the main argument for the general direct effect of the common market provisions? In essence it was this: the Treaty’s historic choice in favour of a gradual implementation during the transitional period did not make them ‘conditional’ provisions after the expiry of that period. Thus, even if, for example, the freedoms of establishment and services expressly referred to ‘the framework of the [legislative] provisions set out below’, that condition had only applied during(!) the transitional period; and once that period had ended, direct effect was to automatically follow. ⁷⁰ To the Court, this solution textually derived from the constitutional effect of ex-Article 8 EEC, ‘according to which the expiry of the transitional period shall constitute the latest date by which all the rules laid down must enter into force and all the measures required for establishing the common market must be implemented’. ⁷¹

After the transitional period, then, the fundamental freedoms had become unconditional norms; yet what about their also being clear and unqualified? To better justify the second element of the classic direct effect test, the Court now expressly linked all fundamental freedoms to a discrimination rationale; and here especially to ex-Article 7 EEC. ⁷² Indeed, according to the Court, it was that provision that ex-Article 52 EEC implemented ‘in the special sphere of the right of establishment’. ⁷³ And, as regards the provisions on the freedom of services, the Court similarly held that they could have direct effect ‘in so far as they seek to abolish any discrimination against a person providing a service by reason of his nationality’. ⁷⁴ This constitutional link between ex-Article 7 EEC and the scope of the fundamental freedoms was also judicially confirmed for the free movement of workers; ⁷⁵ and a discrimination rationale equally arose under the free movement of goods. ⁷⁶ The Court had linked direct effect to discrimination, because only a classic discrimination test was ‘clear’ enough to be applied by the national judiciary. ⁷⁷

Following the ‘1974’ judgments of Reyners, Van Binsbergen, Walrave (and Dassonville), all discriminatory measures falling within these freedoms were, after the end of the transitional period, directly prohibited by the Rome Treaty as part of its negative integration sphere (‘law’). ⁷⁸ From now on, positive integration (‘politics’) could, therefore, concentrate on the harmonisation of those national differences that also posed obstacles to the creation of the common market. ⁷⁹ These obstacles could not be caught by a discrimination rationale, because the latter logically presupposed that it was ‘possible to hold one and the same subject responsible for the prohibited discrimination’. ⁸⁰ And since ex-Article 7 EEC was ‘not concerned with any disparities in treatment or the distortions which may result. . . from divergences existing between the laws of the various Member States’, ⁸¹ these disparities could not be addressed by negative integration; they had to be addressed, instead, by positive integration – to which we will now turn.

B. Positive integration: the direct effect of Community directives

After the transitional period, the relationship between the sphere of negative integration (‘law’) and the sphere of positive integration (‘politics’) appeared to be this: the former was to remove discriminatory national measures, whereas positive harmonisation was to remove obstacles to trade arising from disparities between non-discriminatory national laws. This constitutional division of labour can, for example, be seen in Directive 70/50; ⁸² and it was taken to mean, in particular, that ‘a product which has been manufactured in conformity with French standards could not be sold in Germany because the German authorities apply different standards’. ⁸³ Here, it was the political process – not the Court – that was charged to remove ‘obstacles to international trade resulting from differences between national legislative and administrative provisions’. ⁸⁴

To pursue this positive integration, the Rome Treaty gave a number of legislative competences to the Community. ⁸⁵ Yet for only one fundamental freedom had it permitted the adoption of ‘regulations’ – the supranational instrument expressly allowing for direct effects in the national legal orders. ⁸⁶ By contrast, almost all common market competences envisaged the adoption of ‘directives’. Ex-Article 100 EEC, to mention just one example, thus stated: ‘The Council shall, acting unanimously on a proposal from the Commission, issue directives for the approximation of such provisions laid down by law, regulation or administrative action in Member States as directly affect the establishment or functioning of the common market’.

When compared to negative integration, positive harmonisation thus seemed to generally suffer from a decisional and a normative weakness. For not only did the adoption of secondary Community law require, especially after the Luxembourg Compromise, unanimity in the Council, it also appeared to lack direct effect because directives were, like international law, only binding ‘on’ Member States and solely ‘as to the result to be achieved’. ⁸⁷ On the other hand, positive integration seemed to be wider in scope than negative integration because its competences could, from the start, enter into areas into which the fundamental freedoms – confined to a discrimination rationale – could not; and the Luxembourg Compromise further reinforced this competence asymmetry in favour of positive (!) integration, when the Member States agreed to dramatically expand the Community’s general competences via ex-Article 100 and 235 EEC. ⁸⁸

In a parallel movement, the Court nonetheless quickly decided to strengthen the normative qualities of directives too. The Court indeed discovered their (vertical) direct effects as early as 1970, ⁸⁹ and, more famously again, in 1974. In Van Duyn, ⁹⁰ a Dutch secretary’s entry into the United Kingdom had been halted on public policy grounds. In an effort to harmonise such national derogations, the Community had previously adopted a directive that technically outlawed the British restriction; ⁹¹ yet the United Kingdom had not implemented it into national law. The general question therefore arose whether directives could – contrary to the express wording of ex-Article 189 [3] EEC – have direct effects. And the Van Duyn Court famously found that this was indeed possible:

[B]y virtue of the provisions of [ex-]Article 189 regulations are directly applicable and, consequently, may by their very nature have direct effects, it does not follow from this that other categories of acts mentioned in that Article can never have similar effects. It would be incompatible with the binding effect attributed to a directive by [ex-]Article 189 to exclude, in principle, the possibility that the obligation which it imposes may be invoked by those concerned. In particular, where the Community authorities have, by directive, imposed on Member States the obligation to pursue a particular course of conduct, the useful effect of such an act would be weakened if the individuals were prevented from relying on it before their national courts and if the latter were prevented from taking it into consideration as an element of Community law. ⁹²

Like regulations, directives could thus have direct effects; and this even despite their being binding only ‘as to the result to be achieved’. For the Court had simultaneously clarified that directives were not confined to establishing mere ‘aims’ or ‘directions’; they could, instead, establish ‘exhaustive’ or ‘complete’ harmonisation wherever this was necessary. ⁹³ This federalist pre-emptive effect shall, however, not interest us here; ⁹⁴ rather, it is a more fundamental question: why did the Court supranationalise the directive in the early 1970s?

With many Member States fiercely opposing this normative development from the start and for a long time, ⁹⁵ the constitutional impulse must have come from within the normative sphere itself. Two reasons here come to mind. First and foremost, if state-addressed Treaty provisions – like the fundamental freedoms – could have direct effects after their ‘implementation period’ had ended in 1970, the same logic could of course apply to other state-addressed norms – such as directives – once their implementation period had passed.

But secondly, and more importantly: without the Court giving direct effect to directives, a normative asymmetry between primary and secondary law would have arisen. It would have meant that an individual could directly rely on her abstract Treaty-based freedom (negative integration); yet, ironically, she would not have been able to rely on secondary law that concretised her rights under that freedom (positive integration). ⁹⁶ The simultaneous announcement of the direct effect of primary law (Treaty) and secondary law (directives) in Van Duyn is therefore no coincidence. It was, on the contrary, a manifestation of a self-reinforcing normative supranationalism in which the need for normative symmetry between negative and positive integration was the driving force behind this development!

4. Moving into trouble: the ‘deficient’ legislator and the Cassis revolution

How successful would the Community’s positive integration method be? The extensive harmonization programme of the young Community had, as regards the free movement of goods for example, been reflected in its 1969 ‘General Programme for the elimination of technical barriers to trade arising from disparate national regulations’. ⁹⁷ Yet by 1973, only a third of the envisaged directives had been successfully adopted; ⁹⁸ and a supplementary programme needed to quickly approve a new timetable. That new timetable slipped again, ⁹⁹ but worse: the already low legislative output was given a regressive momentum by the neo-protectionist Member States after the 1973 oil crisis. By 1978, the Commission was exasperated: the Community was ‘adopting fewer and fewer directives each year’! ¹⁰⁰ The decisional intergovernmentalism (‘politics’) within the Community had created a major constitutional crisis: the apparent ‘equilibrium’ between normative and decisional supranationalism had caused a major ‘disequilibrium’ in the Community system!

How was this systemic crisis to be resolved? In order to reinvigorate the Community’s decision-making process, the European Council called, as early as 1974, on the Member States to renounce the Luxembourg Compromise. ¹⁰¹ Yet the Member States, especially the newer ones, continued to fiercely support it. ¹⁰² This only changed in 1986 with the Single European Act (SEA) that reintroduced majority voting a decade later. But, alas, what were the triggers and drivers behind this return to decisional supranationalism? Some have pointed to a convergence of economic interests amongst the Member States (Moravcsik); ¹⁰³ others have emphasised structural (Keohane), ¹⁰⁴ or ideational aspects (Parson). ¹⁰⁵ This section, by contrast, wants to explore the power and potential of a legal driver behind the SEA: the Cassis de Dijon judgment.

A. Expanding negative integration: the Cassis de Dijon wager

Cassis de Dijon is the most important case in all internal market law – if not in all European Union law. ¹⁰⁶ It concerned a German measure that had fixed the minimum alcohol strength of liqueurs to 25 per cent. Formally, the German law thereby applied equally to foreign and domestic goods and was therefore not discriminatory; yet because of it, the French liqueur Cassis de Dijon could not be sold in Germany, as its alcoholic strength was significantly lower. The resulting trade obstacle within the common market had arisen from a disparity between French and German law; and such disparities had traditionally fallen, as we saw in Section 3, within the sphere of positive integration (‘politics’). The Cassis Court however famously rejected this traditional logic in a legendary judgment, which now held:

Obstacles to movement within the Community resulting from disparities between national laws relating to the marketing of the products in question must be accepted in so far as those provisions may be recognized as being necessary in order to satisfy mandatory requirements . . . [T]he requirements relating to the minimum alcohol content of alcoholic beverages do not serve a purpose which is in the general interest and such as to take precedence over the requirements of the free movement of goods, which constitutes one of the fundamental rules of the Community . . . There is therefore no valid reason why, provided that they have been lawfully produced and marketed in one of the Member States, alcoholic beverages should not be introduced into any other Member State; the sale of such products may not be subject to a legal prohibition on the marketing of beverages with an alcohol content lower than the limit set by the national rules. ¹⁰⁷

The judgment profoundly re-structured the constitutional division of labour between positive integration (‘politics’) and negative integration (‘law’), discussed above. From now on, a different constitutional logic applied: unless there was a general public interest that the host State could invoke, the latter was not entitled to impose its own domestic laws on imported goods. This new legal presumption has become known as the ‘principle of mutual recognition’. It meant that Member States must, in principle, mutually recognise each other’s national legislation; and this was, from now on, a constitutional principle of negative integration. ¹⁰⁸

What was the revolutionary idea behind Cassis? Its rationale was as simple as it was radical: so long as the Member States could not agree on positive integration in the Council, the Court could take over and judicially handle all legal conflicts between the home and the host state through negative integration. And yet: it would be a serious mistake to equate the Cassis revolution with negative integration (‘law’) triumphing over positive integration (‘politics’). For let us return to the Cassis formula, quoted above, but now starting from the previous paragraph:

In the absence of common rules relating to the production and marketing of alcohol – a proposal for a regulation submitted to the Council by the Commission on 7 December 1976 (Official Journal C 309, p 2) not yet having received the Council’s approval — it is for the Member States to regulate all matters relating to the production and marketing of alcohol and alcoholic beverages on their own territory. Obstacles to movement within the Community resulting from disparities between national laws relating to the marketing of the products in question must [however only] be accepted in so far as those provisions may be recognized as being necessary in order to satisfy mandatory requirements . . . ¹⁰⁹

The European Court here expressly lamented the ‘absence of common rules’, adopted by the Community legislator, and unmistakably pointed the finger to the Council for having failed to reach a political compromise through the decisional process. And this judicial lament is the critical point: the Community legislator could have harmonised the matter, but pending political agreement in the Council, integration-through-law now provided a second and alternative solution to integration-through-legislation. The expansion of negative integration (‘law’) henceforth offered a fallback solution that was, in the future, to nudge reluctant Member States into finding a legislative compromise in the positive integration sphere (‘politics’). ¹¹⁰ After Cassis, one could say, positive integration was to take place in the shadow of negative integration.

This dialectic relationship between negative and positive integration was confirmed, a few years later, in Polydor. ¹¹¹ The Community had concluded a free trade agreement with a (then) third state, whose formal text clearly reproduced ex-Articles 30 and 36 EEC; and yet the Court unambiguously refused to project its Cassis-jurisprudence into the external realm. Why? According to the Court, this was because of the absence of positive integration competences under international law:

The considerations which led to that interpretation of [ex-] Articles 30 and 36 of the Treaty do not apply in the context of the relations between the Community and [a third state] as defined by the Agreement. It is apparent from an examination of the Agreement that although it makes provision for the unconditional abolition of certain restrictions on trade between the Community and [the third state], such as quantitative restrictions and measures having equivalent effect, it does not have the same purpose as the EEC Treaty . . . In the present case such a distinction is all the more necessary inasmuch as the instruments which the Community has at its disposal in order to achieve the uniform application of Community law and the progressive abolition of legislative disparities within the common market have no equivalent in the context of the relations between the Community and [the third state]. ¹¹²

For the Court, the critical factor in choosing between the ‘international’ and the ‘Cassis’ interpretation of ‘Article 30 EEC’ was therefore whether a trade rule was embedded (or not) in an institutional context that allowed for positive integration. And since, in this context, the Community’s legislative powers could not extend to harmonising third-country goods, there could be no principle of mutual recognition. ¹¹³ This systemic nexus between negative and positive integration can be found in other jurisprudential contexts too; ¹¹⁴ and this close connection seems to suggest this: without the possibility of positive integration in the Community system, there would have been no Cassis, yet, as we shall see in the next section, without Cassis, there may have also been no future to positive integration!

B. The principle of mutual recognition: from Cassis to the SEA

The Cassis judgment provoked two polarised reactions. The Commission instantly put forward a generalisation of mutual recognition in the negative integration sphere: ‘Any product lawfully produced and marketed in one Member State must, in principle, be admitted to the market of any other Member State’; ¹¹⁵ and this, in particular, ought to mean that ‘harmonisation will henceforth have to be directed mainly at national laws . . . which are admissible under the criteria set by the Court’. ¹¹⁶ For most Member States, on the other hand, such a radical generalisation of mutual recognition was utterly unacceptable. ¹¹⁷

In between both views lay the Court that, in the early 1980s, carefully consolidated and refined its Cassis jurisprudence. ¹¹⁸ And, importantly, once its jurisprudence had settled, the principle of mutual recognition had become a legal ‘fact’ that the Member States would have to accept as a general constitutional principle of European law. It was, as a legal fact, picked up by the Council in 1983 and 1984; ¹¹⁹ and the Dooge Committee subsequently endorsed it in 1984 and 1985. ¹²⁰ The most significant push however came in the form of the 1985 White Paper on ‘Completing the Internal Market’, ¹²¹ which would form the bedrock of the 1986 Single European Act. Here we read:

The harmonisation approach has been the cornerstone of Community action in the first 25 years and has produced unprecedented progress in the creation of common rules on a Community-wide basis. However, over the years, a number of shortcomings have been identified and it is clear that a genuine common market cannot be realised by 1992 if the Community relies exclusively on [ex-]Article 100 of the EEC Treaty… In principle, therefore, given the Council’s recognition (Conclusions on Standardization, 16 July 1984) of the essential equivalence of the objectives of national legislation, mutual recognition could be an effective [new] strategy for bringing about a common market in a trading sense… Following the rulings of the Court of Justice, both the European Parliament and the Dooge Committee have stressed the principle that goods lawfully manufactured and marketed in one Member State must be allowed free entry into other Member States. ¹²²

With the 1985 White Paper, the Commission now pushed the principle of mutual recognition beyond technical standards, and even beyond goods. It indeed went unashamedly to the very centre of a ‘re-constituted’ common market. For ‘[w]hat is true for goods, is also true for services and for people’, because ‘[i]f a Community citizen or a company meets the requirements for its activity in one Member State, there should be no valid reason why those citizens or companies should not exercise their economic activities also in other parts of the Community’. ¹²³

But why would the Member States accept such a dramatic expansion of the Cassis revolution in areas in which the Court had not yet juridically confirmed the principle of mutual recognition? The claim of a sudden convergence of national economic interests here significantly underplays, in my view, the power of political ‘ideas’ and legal ‘facts’. For the 1986 SEA was a political project in the footsteps of the 1957 Rome Treaty. ¹²⁴ Like the latter, its principal aim was thus the creation of ‘a genuine political entity among European States’. ¹²⁵ Yet the road to this ‘new political utopia’ was, once more, economic integration, with all the major Member States – apart from Britain ¹²⁶ – willing to ‘play’ the ‘market’ for supranational ‘political’ reforms, like the reintroduction of qualified majority voting and the rise of the European Parliament. ¹²⁷

But let us take one step back: how, concretely, was the SEA to implement the White Paper’s commitment to completing the internal market? Its novel ex-Article 8a EEC here re-introduced a ‘transitional’ period set to end in ‘1992’. ¹²⁸ Its most important innovation was, however, to end the Luxembourg Compromise, at least for the internal market. ¹²⁹ The central provision in that context was, undoubtedly, ex-Article 100a EEC (today Article 114 TFEU). While narrower in scope than ex-Article 100 EEC, ¹³⁰ it now simultaneously strengthened positive integration with regard to both decisional and normative supranationalism. For apart from allowing for qualified majority voting, the Community legislator was also no longer confined to directives. And almost forgotten today, the provision was, thirdly, strengthened by ex-Article 100b EEC (today deleted), which allowed ex-Article 100a to be used, once the new transitional period had passed, to ‘decide that the provisions in force in a Member State must be recognized as being equivalent to those applied by another Member State’.

Ex-Article 100b here embodied a Cassis de Dijon wager writ large. For even if it did not envisage the automatic mutual recognition of all unharmonized national laws by 1992, as had been originally suggested by the Commission, ¹³¹ it nonetheless allowed the Community legislator to do so in one single legislative act – and one to be adopted by a qualified majority of Member States. If sectoral harmonisation within the sphere of positive integration was thus not forthcoming, the Community could – under the conditions set out in ex-Article 100a – complete the internal market by simply generalising the principle of mutual recognition! This threat soon re-galvanised the Community legislator – a legislative renaissance that will be explored in the next section.

5. ‘Integration-through-legislation’: the rise of positive integration

The dramatic rise of European legislation following the SEA has been extraordinary: in the course of two decades, European integration came to re-embrace positive integration; and, methodologically, this also signalled a new ‘integration-through-legislation’ (‘politics’) approach. ¹³² For European legislation was not made by the European Court but by the political organs of the Union deliberating and deciding on which positive rules were to govern Europe. The normative advantages of legislative rules over judicial ones are thereby manifold. ¹³³ Democratically, the legislature enjoys a more direct representational link to the citizens; and, functionally, legislative rules tend to offer clearer and more precise solutions when compared to the often rather casuistic law offered by the judiciary. It is for these reasons that courts ought to defer to legislation; yet would this deference also be shown by the European Court? Let us explore this question after quickly showcasing the spectacular rise of European legislation after 1987.

A. The Cassis wager redeemed: three (plus one) legislative illustrations

Would the Cassis wager, presented in Section 4, ultimately be redeemed? Nothing tells this story better than the fate of the (draft) legislation mentioned in Cassis itself. ¹³⁴ Soon after Cassis, the Commission had re-started working on a legislative proposal ‘laying down general rules on the definition, description and presentation of spirituous beverages’ that was formally published in the Official Journal in 1982. ¹³⁵ Yet the latter, again, remained with the Council for years; ¹³⁶ and only following the SEA’s re-introduction of qualified majority voting could the Council finally adopt a regulation laying down general rules on the definition, description and presentation of spirit drinks in 1989. ¹³⁷ Based, inter alia, on Article 100a EEC (now Article 114 TFEU) – the new internal market competence introduced by the Single European Act, the legislation now offered detailed rules on the definition of ‘liqueur’ for the entire internal market.

This Cassis dynamic of negative integration ‘provoking’ positive integration has, importantly, not been confined to product-related rules only. It can also be seen in the context of so-called selling arrangements. For after a judicial push to maximise negative integration in the Sunday Trading cases, ¹³⁸ the Union legislator quickly pushed back to achieve one of its most important social policy measures: the Working Time Directive. Following a 1990 Commission proposal that had creatively used the post-SEA social policy provisions, ¹³⁹ the Directive comprehensively set out weekly rest and holiday periods for workers. ¹⁴⁰ It was immediately challenged by a pro-market United Kingdom – a challenge that was, however, generally rejected by the Court acknowledging a wide margin of discretion that the EU legislature would enjoy in socio-economic matters. ¹⁴¹

A third illustration may, finally, be cited from outside the free movement of goods. Here, after the Iberian accession in 1986, the Court had – myopically – decided that the EU freedom of services would entail the right of foreign companies to post workers into a host state. ¹⁴² The judgment, unsurprisingly, raised grave concerns about social dumping in Member States with high levels of social welfare and employment protection, especially France. And within a year, the Delors Commission, strongly pressed by the European trade unions, produced a legislative proposal for the posting of workers. ¹⁴³ It took another five long years before a political compromise was reached, yet the resulting 1996 Posted Workers Directive now required host Member States to extend their core labour laws to foreign posted-workers, ¹⁴⁴ while it also allowed them – or so it appeared – to additionally impose stricter host state rules more favourable to workers. ¹⁴⁵

With these three illustrations of European re-regulation in mind, it is also important to note that EU legislation is not just about the harmonisation or the coordination of national laws in a cross-border situation. Like in the past, it will also often liberalise intra-Union trade by removing national barriers. EU legislation here functionally ‘replaces’ the EU Treaties’ fundamental freedoms; and it is the rise of such free movement legislation that explains, to a great extent, the dramatic decline in fundamental-freedom-jurisprudence in recent years, ¹⁴⁶ as the Court’s judgments today primarily relate to Union legislation. ¹⁴⁷

A good example of such ‘negative’ integration through Union legislation is the Services Directive, ¹⁴⁸ which has been said to be ‘the most widely and passionately discussed text of secondary [sic] legislation in the history of the EU’. ¹⁴⁹ Its central pillar had originally been the principle of mutual recognition (‘country of origin’) principle, ¹⁵⁰ but the idea that the home State was the exclusive regulator of a service encountered severe criticism, especially from the European Parliament. ¹⁵¹ The final version of the Services Directive has therefore come to liberalise the internal market in a less absolute manner. ¹⁵² But the general inclusion of mutual recognition clauses in Union legislation following the Single European Act is no coincidence. Instead, it forms part of a shift from total to minimum harmonisation. ¹⁵³ For under the minimum harmonisation model, the Member States continue to be allowed to maintain their stricter national standards; yet in order to guarantee free movement, the Union now insists on the mutual recognition of all Union-compliant national standards. ¹⁵⁴

B. The place of positive integration: the legislative priority rule and the Court

With internal market legislation (‘politics’) increasingly setting out the concrete balance between liberalisation and re-regulation, what is its relationship to negative integration and the EU Court (‘law’)? In other words, how has the Court dealt with the ‘presence of Union legislation’: would it defer to the concrete balance struck in the Union legislative act; or would it give normative priority to the more abstract EU fundamental freedoms set out in the EU Treaties?

The judicial answer to this question has recently become known as the ‘legislative priority rule’, ¹⁵⁵ according to which the Court gives, as a rule, applicative priority to EU legislation. A classic formulation here states that ‘where a matter is regulated in a harmonised manner at [Union] level, any national measure relating thereto must be assessed in the light of the provisions of that harmonising measure and not of [the fundamental freedoms] of the Treaty’. ¹⁵⁶ The Court will thus ‘suspend’ the application of the EU Treaty’s fundamental freedoms in the presence of – exhaustive – Union legislation. ¹⁵⁷

This legislative-priority-rule has a vital constitutional corollary. For when translating the EU fundamental freedoms into secondary law, the Union legislator enjoys a wide margin of discretion; and the resulting political freedom means that the Union can adopt collective solutions that the Member States individually could not. In Bauhuis, ¹⁵⁸ for example, Union legislation was challenged on the ground that the compulsory health checks it created when animals were exported to other Member States violated the EU Treaties, including Article 34 TFEU (ex-Article 30 EEC). Legally, there was no doubt that national border inspections were measures prohibited by this provision. Yet the Court found that a different normative standard would apply to the Union legislator.

Why? Because EU legislation would, unlike national legislation, establish a Union-wide standard that furthered the ‘unity of the market’, and it followed that the ‘criteria to appreciate the legality of national and [Union] measures with respect to their effects on the Common Market cannot be the same’. ¹⁵⁹ This fundamental difference in the normative status and role of EU legislation, when compared to national legislation, indeed stems from the different interests they represent. Unlike the particularistic national interests, represented by the Member States, the EU legislator stands for the general Union interest – an interest that is unlikely to be tainted by a spirit of national protectionism. ¹⁶⁰ The Union legislator will thus be able to legislate in a substantially freer way than national legislators; and it is even permitted to restrict intra-Union trade to the highest degree if it so wishes! ¹⁶¹

But what if the Court of Justice disagrees with such a high regulatory standard? A recent example of this eventuality comes from the 1996 Posted-Workers-Directive. We saw above that the original legislative regime here appeared to have been designed to protect the national labour laws of the host state. Yet in the course of time, the Directive had received a very controversial neoliberal interpretation by the Court, especially in Laval and Commission v Luxembourg. ¹⁶² These cases now sternly limited the ability of host states, and their trade unions, to impose higher social protection standards on posted workers. Could the EU legislator, unhappy with this judicial re-interpretation, subsequently correct the Court by amending the legislation? Spurred and supported by the Commission, ¹⁶³ the Union legislator clearly believed so:

According to Article 3 of the Treaty on European Union, the Union is to promote social justice and protection. According to Article 9 TFEU, in defining and implementing its policies and activities, the Union is to take into account requirements linked to the promotion of a high level of employment, the guarantee of adequate social protection, the fight against social exclusion, and a high level of education, training and protection of human health. More than 20 years after its adoption, it has become necessary to assess whether Directive 96/71/EC of the European Parliament and of the Council still strikes the right balance between the need to promote the freedom to provide services and ensure a level playing field on the one hand and the need to protect the rights of posted workers on the other. . . . Ensuring greater protection for workers is necessary to safeguard the freedom to provide, in both the short and the long term, services on a fair basis, in particular by preventing abuse of the rights guaranteed by the Treaties. ¹⁶⁴

The pro-labour legislative amendments were immediately challenged by a number of Eastern European States, two of which ultimately brought annulment proceedings before the Court. ¹⁶⁵ There, the applicants alleged a violation of the underlying EU fundamental freedom, because the amended directive ‘creates restrictions on freedom to provide services that are contrary to Article 56 TFEU’. ¹⁶⁶ The Court however unconditionally confirmed the power of the EU legislator to shape and define the specifics of all fundamental freedoms in the following manner:

[I]n relation to the free movement of goods, persons, services and capital the measures adopted by the EU legislature, whether measures for the harmonisation of legislation of the Member States or measures for the coordination of that legislation, not only have the objective of facilitating the exercise of one of those freedoms, but also seek to ensure, when necessary, the protection of other fundamental interests recognised by the Union which may be affected by that freedom. That is the case, in particular, where, by means of coordination measures seeking to facilitate the freedom to provide services, the EU legislature takes account of the general interest pursued by the various Member States and adopts a level of protection for that interest which seems acceptable in the European Union… With regard to judicial review of compliance with those conditions, the Court has recognised that, in the exercise of the powers conferred on it, the EU legislature must be allowed a broad discretion in areas in which its action involves political, economic and social choices and in which it is called upon to undertake complex assessments and evaluations. ¹⁶⁷

The Court’s strong constitutionality presumption for legislative choices of the Union, especially socio-economic ones, here belies the view that considers the Union legislature (‘politics’) as an agent of the European Court of Justice (‘law’). ¹⁶⁸ For even if the EU legislator will – of course – keep a very keen eye on the Court and often react to it, the legislative priority rule generally suspends the fundamental freedoms wherever the Union legislature has harmonised a subject (in a not totally unreasonable manner). ¹⁶⁹ The Court, today, therefore mostly operates as an ‘ordinary’ court that interprets EU legislation; and the internal market case ‘law’ that results from this interpretation, having the same status as the ‘legislation’ it interprets, also no longer constitutes an absolute barrier to future EU ‘politics’.

6. History meets theory: ‘Integration-through-Law’ revisited

Let us now take stock. Having previously analysed three concrete historical balances between normative and decisional supranationalism as well as negative and positive integration in Sections 2–4, and having specifically explored the normative status of EU legislation in Section 5, what theoretical insights can we abstract from all this historical material? In particular: could we, with Weiler’s equilibrium theory, find an inverse relationship between normative supranationalism (‘law’) and decisional supranationalism (‘politics’)? Or were there, with Scharpf, major asymmetries between negative and positive integration? And what about Grimm’s over-constitutionalisation thesis? Let us revisit and reassess all three famous integration-through-law theories in this final section – starting with the third and youngest integration-through-law variant first.

A. The over-constitutionalisation theory: text of teleology?

Is the internal market ‘over-constitutionalised’? Almost half of the original Rome Treaty covered the internal market, because its establishment and future functioning was its essential aim. (For an idea of how complex and cumbersome many of the original EEC provisions were, take another look at the Appendix!) Yet the relevant Treaty text on the internal market has significantly reduced since then; and after the Cassis revolution, most Treaty-based internal market law has been created out of single constitutional provisions – each for one fundamental freedom. As regards the free movement of goods, this constitutional text has been Article 34 TFEU which contains but 16 words – hardly a number that amounts to textual over-constitutionalisation. On the contrary, and as has been argued elsewhere, the secret behind the enormous success of the internal market provisions lies in their textual underdetermination, which has allowed the Court to ‘transform’ the EU internal market from an international to a federal market order. ¹⁷⁰

Yet what about the EU Treaties’ teleological over-constitutionalisation? The problem with this line of argument, however, is that most modern states will have insisted, in the past, on the creation of a domestic – national – market in which the free movement of goods, persons, services and capital is constitutionally guaranteed. This economic aim is thus not a sui generis objective of the Union alone. It has, for example, been part and parcel of the US American constitutional order, where the Supreme Court has pressed the ‘Commerce Clause’ even beyond what the European Court has done so far. True, the US Constitution pursued, since its inception, many more substantive aims; but the European Union has – over time – also considerably diversified its constitutional aims, particularly since the Lisbon Treaty. When compared to ex-Article 2 EEC, a look into Article 3 TEU thus clearly shows that ‘the very purpose and telos of the European Union has enlarged’. ¹⁷¹ The Union constitutional order is today, as regards its objectives or ‘telos’, an increasingly ‘open’ political order with the EU legislator enjoying a wide margin of discretion to pursue the Union’s diverse and variegated aims and competences.

What, however, explains the lengthy text of the 1957 EEC Treaty and its enormous textual increase ever since? It is hard to follow Grimm’s claim that the original ‘over-constitutionalisation’ was the attempt to ‘immunise[] the Commission and particularly the [European Court of Justice]’. ¹⁷² For was it not the Member States themselves that had – as collective Masters of the Treaties – deliberately chosen to textually fix the exact way in which European integration was to proceed? This change of perspective also better explains why subsequent Treaty amendments would add ever more EU constitutional law. For it is against the background of significant informal changes of the EU Treaties – both in the negative and positive integration sphere – that the dynamics and forces behind the exponential growth of EU constitutional law ought to be understood.

The Union’s general competences, for example, had been exploited to informally expand the Union’s legislative sphere; ¹⁷³ and the emergence of textually complex complementary competences and the inclusion of ever ‘wordier’ Protocols – both constitutionally designed to harness the Union – here constitute clear manifestations of the Member States trying to ‘immunise’ themselves – not the Commission or the Court – against all organic change within the Union legal order. Because more text means more precision; and more legal precision means less informal freedom for the Union. The post-1986 over-constitutionalisation of the EU Treaties thus appears to be a reaction and result of the historical evolution of the Union. It is the consequence – not the cause – of an integration-through-law approach that Grimm’s theory simply cannot explain.

Let us therefore concentrate on the first and second theory in the remainder of this Section.

B. The equilibrium theory: inverting the inverse relationship?

This cannot be the place to subject Weiler’s integration-through-law theory to a comprehensive ‘re-reading’. Section 2 has, however, already hinted at a serious reservation about the revolutionary nature of the doctrine of direct effect; ¹⁷⁴ but for the purposes of this section, the core paradox within Weiler’s equilibrium theory is this: what does the Court’s choice in favour of the direct effect of Treaty articles have to do with the decisional processes in the Union? Put more concretely: if ex-Article 12 EEC was never meant to be implemented by Union legislation, how can the Luxembourg Compromise be said to balance the normative supranationalisation of the provision after Van Gend?

Weiler’s equilibrium theory could, at best, make sense as a theory for secondary Union law (positive integration). ¹⁷⁵ Yet with regulations constitutionally designed to be directly applicable in the national legal orders from the start, all here depends on directives, which are hardly discussed by Weiler’s theory. If a bit more jurisprudential work had here been done, ¹⁷⁶ Weiler might have also found that all important judgments on the direct effect of Treaty articles were given after his foundational period had ended. This, of course, significantly relaxes the associative connection between ‘1963’ (Van Gend) and ‘1964’ (Costa), on the one hand, and ‘1965’ (empty chair crisis) and ‘1966’ (Luxembourg Compromise), on the other. And yet: could one not still argue, as regards the internal market, that the decline in decisional supranationalism after the Luxembourg Compromise was compensated by the rise of normative supranationalism in ‘1974’ (Reyners, Van Binsbergen, Van Duyn, Dassonville)? This is not Weiler’s thesis, but a different one, later developed by Paul Craig; yet even this new equilibrium theory suffers, in my view, from explanatory difficulties of its own. ¹⁷⁷

But be that as it may, Weiler’s equilibrium thesis is badly misconceived for a second reason. For Cassis de Dijon – a case that plays no role in his grand theory of European integration– simply cannot be understood in normative supranational terms. ¹⁷⁸ The essence of Cassis is not the normative quality of Article 34 TFEU; it is rather the revolutionary choice to allow the (intergovernmental) ‘politics’ in the EU legislature to be replaced by the (supranational) ‘law’ of mutual recognition. And it is this move that goes directly against Weiler’s equilibrium theory because by potentially removing the political need to get the actual consent of the Member States in the Council to establish internal market law, the Court assumes a role within the decisional supranational sphere. Supranational ‘law’ is here not counterbalanced by intergovernmental ‘politics’; the former stands in for the latter!

Furthermore, if one believes – as I do – that the Cassis Court’s ultimate intention was to push the EU legislature into action within the positive integration sphere, Weiler’s entire theory needs to be put – as Marx did with Hegel – on its head. Indeed, after 1979, Cassis inverted the inverse relationship that Weiler claims to have discovered between the legal and the political spheres. For it is, as we saw in Section 4, the imbalance between negative integration (‘law’) and positive integration (‘politics’), created by the Cassis jurisprudence, that is – seven years later – rebalanced by more decisional supranationalism in the political sphere. ¹⁷⁹ More supranational law, more supranational politics! And Weiler’s theory here not only misses the link between Cassis and the Single European Act; it also misses the positive correlation between the rise of majority voting in the Council and the rise of the European Parliament.

Let me explain this last point by means of a small, but hopefully revealing, excursion.

The original Union ‘legislator’ had, of course, been the Council alone; and after the Luxembourg Compromise, this was a Council in which each and every Member State could potentially veto each and every legislative act. After the SEA, the Union had however returned to decisional supranationalism with its core provision now allowing the Council ‘acting by a qualified majority on a proposal from the Commission in cooperation with the European Parliament’ to adopt EU legislation. ¹⁸⁰ This new legislative procedure had – nota bene – not only abandoned unanimity voting in the Council; it now also required the ‘cooperation’ of another supranational institution: the European Parliament. ¹⁸¹ Is there, then, perhaps a constitutional link between the reintroduction of QMV in the Council and the rise of the supranational European Parliament after the SEA?

The original role of the European Parliament under the 1957 Rome Treaty had been minimalist: it was only to be consulted. It had gained some budgetary powers in the early 1970s, when the Community shifted to a system of own resources. (We saw in Section 2 that this reform had, at first, been blocked by de Gaulle’s empty chair policy, yet after his departure in 1969, this change was swiftly implemented; and this, in turn, triggered Parliament’s transformation into a directly elected organ. ¹⁸² ) With the Single European Act, ¹⁸³ Parliament was suddenly given legislative powers under those exact competences in which qualified majority voting had been reintroduced. Why? Three potential answers may be suggested here.

A first answer draws on the powerful ‘supranational’ idealism that many (though not all) Member States still shared. ¹⁸⁴ For the Dooge Committee, for example, the rise of the European Parliament was a constitutional value in itself because it was ‘a guarantor of democracy in the European system’; ¹⁸⁵ and when preparing the SEA, it therefore had clear words: ‘A parliament elected by universal suffrage cannot, if the principles of democracy are logically applied, continue to be restricted to a consultative role or having cognizance of only a minor part of [Union] expenditure’. Parliament’s role had perforce to be enhanced, especially ‘by effective participation in legislative power’. ¹⁸⁶

A second – complementary – answer points again to the logic of ‘compensatory constitutionalism’. ¹⁸⁷ (This logic had, it has been argued, already operated in favour of the European Parliament when the Community moved towards a supranational financing system based on own resources in the 1970s.) The post-SEA move from unanimity to majority voting in the Council here, ultimately, meant a decline in national parliamentary control over European integration, which would need to be ‘re-balanced’ by the rise in European parliamentary control. ¹⁸⁸

A third answer, finally, explains the inclusion of the European Parliament by means of an intergovernmental logic. For under the SEA-introduced cooperation procedure, as employed in the original ex-Article 100a EEC, any parliamentary disagreement with the Council allowed the latter to return to unanimity voting – something that the Member States might have liked. Subsequent Treaty reforms have, however, weakened this third argument significantly. For already the 1992 Maastricht Treaty replaced most instances in which the cooperation procedure had been used with a new ‘co-decision’ procedure under which this intergovernmentalist logic would no longer apply. ¹⁸⁹

But be that as it may. The fundamental point this small excursus wished to make is this: whatever one makes of these three alternative explanations of the post-SEA rise of the European Parliament, the crucial point is that there is always a positive correlation between the rise of decisional supranationalism in the Council and the rise of more decisional supranationalism through the legislative involvement of the European Parliament. More decisional supranationalism triggers more decisional supranationalism! And a similar self-reinforcing logic could, as we saw in Section 3, also be found in the normative supranationalism sphere. For once the Court had generalised the direct effect of the EU Treaty freedoms in 1974, directives were immediately found to have direct effects too. Why? Because it would have been legally inconvenient, if not outright impossible, to deny the direct effect of directives when they were meant to concretise and complement directly effective primary law.

In conclusion, then: the internal dynamics within the normative sphere (‘law’) and the decisional sphere (‘politics’) appear to have been the principal driving forces behind their respective constitutional development; and when there is an interaction between both spheres, like the interaction between Cassis and the SEA, the relationship between ‘law’ and ‘politics’ seems also self-reinforcing. There never was an ‘equilibrium’ between normative and decisional supranationalism in the zero-sum inverse logic suggested by Weiler – neither in the foundational period of EU integration, nor ever hereafter. ¹⁹⁰ The equilibrium theory is a serious historical mistake; and yet it has become a ‘myth’ that keeps on ensnaring! As a historical myth, however, it was nonetheless important, because it emphased the role and relevancy of ‘law’ to a generation of political scientists co-studying the process of European integration. ¹⁹¹

C. The asymmetry theory: decisional, normative, competence?

What about Scharpf’s asymmetry theory? From among the three integration-through-law theories revisited in this article, Scharpf offers the best account of the historical picture, although some of his strong assumptions have recently and righty come to be questioned. ¹⁹²

There is no doubt that the doctrines of direct effect and primacy have reinforced the normative character of the fundamental freedoms (negative integration); yet it is – in my view – but a slight of hand to claim that the rise of normative supranationalism expanded the sphere of negative integration. For as Section 2 has shown, the direct effect of the EU fundamental freedoms was originally limited to provisions that were meant to be self-executing from the start; and even after the transitional period had ended, the Court had, at first, only confirmed the direct effect of negative integration in relation to the principle of non-discrimination. Due to this ‘international’ scope, the direct effect of the EU fundamental freedoms however only formally strengthened a substantive GATT standard that, at least for goods, already bound the Member States under international law.

Van Gend therefore did not imply Cassis: the rise of normative supranationalism did not, as such, lead to an expansion of the sphere of negative integration. This expansion only happened with and after the Cassis revolution, when negative integration enters into a sphere originally reserved to positive integration; and once there is this overlap, Scharpf is correct to highlight the decisional asymmetries between negative and positive integration because the supranational decision-making power of the Court enjoys a clear advantage over any intergovernmental decision-making within the EU legislature.

This decisional asymmetry has, nonetheless, become much less pronounced since the SEA and the subsequent rise of QMV. ¹⁹³ Scharpf indeed significantly undervalues the rise of legislation, discussed in Section 5, which today – more or less – ubiquitously and comprehensively sets the balance between deregulation and reregulation in the internal market. ¹⁹⁴ The Court can, of course, try to alter that legislative balance in ways unintended or unwanted by the EU legislator, yet the latter can also correct the Court interpreting its legislation if it wants to. This might not always be easy because of the need to build political coalitions in the Parliament and the Council, but to speak of a normative asymmetry in favour of the Court is misleading in light of the legislative priority rule discussed above. ¹⁹⁵

Nor is there – pace Scharpf – a competence asymmetry disadvantaging positive integration. For while it is true that the rise of complementary competences after the SEA and ‘1992’ has come to severely limit – and over-constitutionalise – the legislative competences of the Union, the Union’s constitutional ability to use its general competences (Articles 114, 115, 352 TFEU) continues to allow it to legislate on (almost) all matters. ¹⁹⁶ Take, for example, Laval and the right to strike. Here, the Court had judicially limited that right at the national level, while the specific Union competence seemed to expressly exclude any legislative involvement of the Union. ¹⁹⁷ Yet this did not mean that the Union had no competence to legislate in this area. It could and indeed did try to legislative under Article 352 TFEU (‘Monti II Regulation’). ¹⁹⁸ And the key problem here was a decisional asymmetry in the form of an unanimity requirement that prevented the Union from exercising its – existing – legislative competence.

This decisional asymmetry between the Court and the Council is clearest where a single Member State is able to block EU legislation under, say, Articles 115 or 352 TFEU. It can, however, also be of a ‘political’ nature, where majority voting is allowed. For Scharpf is – again – correct to point out that with every EU membership enlargement the political heterogeneity within the EU legislature has increased. (The polarised reactions to Laval and the reform of the Posted-Workers Directive bear witness to that. ¹⁹⁹ ) Yet are these political asymmetries of a structural and permanent nature; and will they thus condemn the Union to a neoliberal Hayekian project? ²⁰⁰

There can be little doubt that the politico-economic interests of the European ‘centre’ and those of the European ‘periphery’ will sometimes be opposed. ²⁰¹ Member States do differ in their structural position within the European economy as well as in their ideological preferences. But why should we assume that these cleavages will be permanent? Is there hope for a broad societal convergence in the long term? I think there is, but in order to achieve that convergence the Union must find better ways to compensate all (temporary) ‘losers’ of European integration, whether they be individual citizens or Member States. ²⁰² This however brings us to a new and crucial asymmetry within the Union: its fiscal asymmetry.

The European Union is a legislative giant on clay fiscal feet. Its budget is – when measured against its legislative tasks – miniscule; ²⁰³ and its fiscal income still predominantly consists – despite what the EU Treaties formally demand – of Member State contributions. But how can this asymmetry between legislative and fiscal integration be rebalanced? One answer here might – Cassis-like – point to the Court. Yet there are inherent limits to any ‘Cassis wager’ in the fiscal sphere. ²⁰⁴ And while there may be other judicial strategies, ²⁰⁵ a more conservative answer always returns to the Member States as unanimous ‘legislators’ (under Article 311 TFEU) or unanimous Treaty-makers (Article 48 TEU). The core problem of fiscal integration, and of European re-distributive ‘politics’ in general, thus comes back to a decisional problem.

In conclusion, then, Scharpf’s thesis about a decisional asymmetry between negative integration (‘law’) and positive integration (‘politics’) continues to hold true, ²⁰⁶ albeit less in its broader formal-legislative than in a more restrictive material-fiscal way. Why would the Member States ever wish to give up their veto powers, especially over fiscal matters, and drop a ‘republican intergovernmentalism’ that celebrates national sovereignty? ²⁰⁷ This question brings me to a fifth and final asymmetry, which is not an asymmetry within the Union but an asymmetry within the Member States. It has been pointedly expressed by the late David Held:

Throughout the nineteenth and twentieth centuries, theorists of democracy have tended to assume a ‘symmetrical’ and ‘congruent’ relationship between political decision-makers and the recipients of political decisions. In fact, symmetry and congruence have often been taken for granted at two crucial points: first, between citizen-voters and the decision-makers whom they are, in principle, able to hold to account; and secondly, between the ‘output’ (decisions, politics and so on) of decision-makers and their constituents – ultimately, ‘the people’ in a delimited territory… But the problem, for defenders and critics of modern democratic systems, is that regional and global interconnectedness contests the traditional national resolutions of the key questions of democratic theory and practice. ²⁰⁸

Economic globalisation, despite all its efficiency gains, has indeed significantly weakened the public powers that most nation states can exercise over their national economy. ²⁰⁹ How can this governmental asymmetry be addressed? If one believes, as I do, that the collective problems humanity faces require collective solutions, then strengthening decisional supranationalism has much to recommend. From its very beginning, the Union was thus meant to ‘govern’ the external pressures of globalisation, ²¹⁰ and it has today, undoubtedly, more governmental capacities than those smaller European states that have been captured by American (or Chinese) corporations and the hegemonic forces of hyper-capitalism. ²¹¹ The future of effective government in Europe thus lies in a ‘republican federalism’ that is capable of exercising public powers at a level most suited to the size of the problem at hand.

7. Conclusion

How has the EU internal market, and with it the original European Union, been integrated in the past; and what were the respective roles of ‘law’ and ‘politics’ in this process? This was the broad question this article wished to explore. We saw above that the relationship between the EU Court and the EU legislature has changed over time. In a first period, discussed in Section 2, the Union legal order gave preference to positive integration, as the common market was to be gradually built over a period of 12 years. After the transitional period had ended, however, negative integration (‘law’) became fully autonomous and, as a consequence, positive integration (‘politics’) came to be seen as operating in a separate sphere – a division of legal labour that was analysed in Section 3. Accordingly, discriminatory national laws fell within the negative integration sphere, whereas non-discriminatory trade obstacles arising from disparities in national laws fell into the positive integration sphere.

This radically changed with Cassis de Dijon – a revolution that was analysed in Section 4. It is with Cassis – not Van Gend or Costa – that the integration-through-law philosophy reaches its climax. ²¹² Yet with Cassis, the idea that normative and decisional supranationalism were in some form of ‘equilibrium’ lies in shatters. ²¹³ For the constitutional dynamic that this case engenders is the very oppositive of what Weiler’s theory had predicted. The supranational judicial sphere (‘law’) here actively helps the European political sphere (‘politics’) to regain supranational agency. The Court (C) helps to sublate national legislation (L) into supranational legislation (L’). Yet through this L-C-L’ logic, Cassis is – dialectically – also the beginning of the end of the integration-through-court paradigm. For by re-galvanising the Union legislator, the judgment ultimately gives rise to a new ‘integration-through-legislation’ paradigm that was discussed in Section 5.

What could, in retrospect, legitimise the Cassis revolution? In the past, some scholars have tried to justify the Court’s judicial activism as ‘democratic’, when viewed from a European perspective. ²¹⁴ This approach is, however, empirically and normatively untenable. ²¹⁵ A second approach has, by contrast, concentrated on ‘liberal’ outputs; yet this approach also suffers from major problems. ²¹⁶ A third legitimation approach has, finally, argued that the Cassis revolution should be justified on ‘republican’ grounds. ²¹⁷ What are the broad outlines of this legitimation strategy? If one accepts, with Section 6, that the nation state has become too small for the global economy and that the ‘spheres of justice’ (Walzer) must be expanded, then the Court’s judicial activism could potentially be seen as assisting in forging a new political community (‘politics’). The integration of the EU market is here viewed as a neo-functionalist means to a neo-federal end. ²¹⁸

This strategy has been very successful in the past. Yet European integration under the banner of the ‘internal market’ has ‘run its course’. ²¹⁹ In the last two decades, the Union has thus been accused of having itself become a neo-liberal tool, because it has not managed to redeem the quintessential promise of all political integration, that is: social integration. The Union urgently needs a ‘European Social Act’ (preceded and supported by a ‘European Fiscal Act’). Why social and fiscal integration? Because the success of modern states in normatively ‘integrating’ their citizens was mainly achieved – the horrors of nationalistic wars aside – through the ‘welfare state’ in which the ‘winners’ and ‘losers’ of economic nationalism were reconciled by considerations of distributive justice. ²²⁰ The Union should follow this path and rediscover – to use Koselleck’s evocative phrase - its ‘futures past’. ²²¹

But in order to get there, the Union must rid itself of all unanimity voting – whether at the constitutional or the legislative level. Only once this outdated international law remnant is dropped, can the decisional and governmental asymmetries that are affecting the Union and its Member States be seriously addressed. The future equilibrium of the Union system indeed may depend on it. For any broad social legitimacy among its citizenry will only ever be re-established if the Member-States-in-Union offer convincing solutions to those everyday problems that are today transnational problems: how to organise the modern economy (work), how to protect the environment (life); and how to reign in transnational private companies and hostile public external actors. All these questions can no longer be decided in splendid national isolation, even if some wish it so.

We can therefore only hope that the Member States remember, as after the Second World War, ²²² that the best way to ‘rescue’ themselves is within a stronger European Union. Yet this is no plea for European integration for European integration’s sake; nor is it pushing market integration for economic efficiency’s sake. This is an argument for effective democracy’s sake: because if democracy refers to the ability to collectively decide on certain outputs, it loses its core meaning when the very ability to positively determine these outputs is lost. How can this ‘sovereign’ ability to collectively choose for Europe and one’s Member State be regained; and what can the future role of ‘law’ here be? The claims that Europe’s ‘messianic’ potential is spent (Weiler); or that the Union is but a ‘Hayekian’ project (Scharpf); or, that it is undemocratic and ‘over-constitutionalised’ (Grimm) tragically deflect from the constructive role that ‘integration-through-law’ can still play in pushing for further political integration. The history of European integration here offers stark lessons; and we should be very critical of any grand theory of European integration that ignores them.