The Normative Impact of Invoking Directives: Casting Light on Direct Effect and the Elusive Distinction between Obligations and Mere Adverse Repercussions


 The main aim of this paper is to cast light on the case law on direct effect of directives, which has remained elusive to both scholars and practitioners. To this end, we first revisit the relevant case law on inverse vertical, horizontal. and triangular disputes to show that the fundamental distinction drawn by the case law is that between ‘direct obligations’ and ‘mere adverse repercussions’. Subsequently, we propose a doctrinal approach to distinguish between ‘direct obligations’ and ‘mere adverse repercussions’ which centres on the impact of invoking a Euorpean Union (EU) directive on the norms governing the dispute. This ‘normative impact theory’ explains all existing case law on the direct effect of directives, and thus aids a better understanding of the concept of imposing obligations on individuals. We compare this theory with other doctrinal theories that have purported to explain the case law, including the well-known distinction between invocabilité de substitution and invocabilité d’exclusion, concluding that the normative impact theory has descriptive and normative advantages over existing approaches. Lastly, we show how the functioning of the preliminary reference procedure has affected the development of the case law on direct effect. We demonstrate that the European Court of Justice (ECJ) applies a presumption that consistent interpretation is capable of remedying incompatibilities between national and EU law. Secondly, we show how the formulation of the preliminary reference can substantially affect, and even confuse, the answer of the ECJ as regards matters of direct effect.


I. Introduction
European Union (EU) law scholars are well acquainted with the lights and shadows of the doctrine of direct effect. 1 Despite Pescatore's well-known claim that direct effect is an 'infant disease' of European law, 2 questions of direct effect continue to arise, particularly in the context of directives. 3 The main aim of this paper is to cast light on a specific aspect of this doctrine: the elusive distinction between obligations and mere adverse repercussions with a view to the invocability of directives.
After many attempts to reconcile the alleged incongruences in the case law on direct effect of directives, and numerous categories and distinctions including not only vertical and horizontal direct effect, but also inverse vertical direct is influenced by the functioning of the preliminary reference procedure (Sections V-VI).
More specifically, this article is structured as follows. First, we introduce the various theories on direct effect, with particular focus on doctrinal explanation of the case law on the horizontal direct effect of directives (Section II). Secondly, we clarify the case law on the invocability of directives, which has remained elusive notwithstanding much doctrinal work in the 1990s and 2000s, by putting the distinction between 'mere adverse repercussions' and 'direct obligations' in the spotlight (Section III). Whilst this analysis includes revisiting some wellknown cases, it is necessary because the existing literature has focused mainly on horizontal direct effect, 9 thus missing the jurisprudential image that emerges from the case law as a whole. Thirdly, we show how the existing theories of the direct effect of directives are incapable of recognising the distinction between direct obligations and mere adverse repercussions (Section IV). Fourthly, we provide a doctrinal construction of this case law, 10 focusing on the 'normative impact' of the invocability of EU directives, which explains how the Court distinguishes mere adverse repercussions from direct obligations (Section V). We conclude that the normative impact theory is descriptively more accurate and normatively more desirable than the existing theories. Fifthly, in Section VI we explain how the functioning of the preliminary reference procedure has affected the development of the understanding of the case law on direct effect, mainly through an implicit presumption of the possibility of consistent interpretation in the absence of a specific question on direct effect, in line with the Court's so-called 'stone-by-stone approach' (Section VI.A), 11 but also through the fact that the Court is always guided-and sometimes potentially taken off course-by the information supplied by the referring national court (Section VI.B). Section VII concludes.

II. Setting the context: direct effect and directives
In Van Gend en Loos, the reasons for the direct effect of Article 12 of the Treaty establishing the European Economic Community (EEC) given by the Court suggested that directives could never be directly effective, since any rights or obligations contained in directives are 'qualified by any reservation on the part of states which would make its implementation conditional upon a positive legislative measure enacted under national law'. 12 On a strict reading of 'unconditionality' as meaning 'without conditions', directives would by definition lack unconditionality.
Nonetheless, the Court's well-known gradual loosening of the conditions of direct effect has allowed it to accept a direct effect of directives as well, while more generally, the increasingly lenient test for direct effect has led commentators to question the relevance of direct effect other than a test of justiciability. 13 These developments need no extensive repetition here. 14 In Reyners, the Court held that Article 52 EEC possessed direct effect, notwithstanding that this provision provided for the gradual aboliton of all barriers to establishment. Thus, while being dependent on further implementation, according to the Court Article 52 EEC had 'at least a partial direct effect in so far as it specifically prohibits discrimination on grounds of nationality'. 15 A similar approach was taken in Defrenne, which concerned an obligation for Member States to gradually abolish all forms of gender discrimination at work by virtue of Article 119 EEC (now 157 of the Treaty on the Functioning of the EU (TFEU)). The Court identified a directly effective core right not to be directly discriminated against. 16 Meanwhile, direct effect was extended to sufficiently clear and unconditional provisions of regulations, 17 decisions, 18 and directives. 19 In Van Duyn, the Court provided three arguments in support of the invocability of unimplemented directives against the state: the binding effect of directives pursuant to Article 288 TFEU, the principle of effectiveness, and the non-exclusion of directives from the preliminary reference procedure. 20 Further, Van Duyn made it clear that the 'unconditionality' criterion should not be given a strict reading, to the extent 12 Van Gend en Loos, EU:C:1963:1. This suggestion is indeed supported by the text of Art. 288 TFEU which appears to deny directives having effects within the Member States' legal orders. 13 S Prechal, Directives in EC Law (Oxford University Press, 2005), 240-1; Pescatore, 'The Doctrine of "Direct Effect"' (n 1), 176-7; R Schütze, European Constitutional Law (Cambridge University Press, 2012), 314. 14 For overviews of the trend from a 'strict' to a 'lenient' test, see eg Schütze (n 1), 268-70; P Craig and G de Búrca, EU Law: Text, Cases, and Materials (6th edn, Oxford University Press, 2015), 184-208; de Witte (n 1). 15 Reyners 2/74, EU:C:1974:68, para. 14. 16 Defrenne 43/75, EU:C:1976:56, paras 21-24. 17 Leonesio 93/71, EU:C:1972:39. 18 Grad 9/70, EU:C:1970:78. 19 Van Duyn 41/74, EU:C:1974:133. 20 Ibid para. 12. In Ratti 148/78, EU:C:1979:110, para 22, the Court added a fourth argument, namely that Member States may not rely on their own failure to implement a directive (the so-called estoppel argument).

Squintani and Lindeboom 22
that limitations of and exceptions to the main rule do not prevent a provision from being unconditional insofar as 'the meaning and exact scope of the provision raise questions of interpretation, these questions can be resolved by the courts'. 21 In light of these developments, Pescatore was the first commentator to observe that direct effect is simply 'the normal state of the law'. 22 In a mature legal order, it is obvious that provisions can be relied upon before courts insofar as they are justiciable, 23 meaning that they can be interpreted and applied by courts. 24 The criteria of sufficient precision and unconditionality embody this requirement. 25 This wide interpretation of the doctrine of direct effect is also supported by judgments such as Becker, 26 VNO, 27 and Kraaijeveld, 28 which conjunctively suggest that justiciability is a necessary and sufficient condition for EU law provisions to be invoked before national courts. In Becker, the Court had seemingly taken the view that-following a strict reading of Van Gend en Loos-there is a link between direct effect and the existence of a subjective right: [W]herever the provisions of a directive appear, as far as their subject-matter is concerned, to be unconditional and sufficiently precise, those provisions may, in the absence of implementing measures adopted within the prescribed period, be relied upon as against any national provision which is incompatible with the directive or in so far as the provisions define rights which individuals are able to assert against the state. 29 While it has indeed been argued that direct effect should be limited to subjective rights, 30 it appears that the word 'or' implies that a subjective right is a sufficient, but not a necessary condition in order for an EU provision to have direct effect. 31 Alternatively, it has been argued that the last part of the cited paragraph is meant to indicate that a subjective right to invoke an EU provision is the consequence of that provision having direct effect. 32 This interpretation has support in the case law as well. Judgments including VNO and Kraaijeveld, in turn, confirm such an 'objective' conceptualization of direct effect-in which justiciability is a necessary and sufficient condition for direct effect-by emphasizing that sufficiently clear and unconditional provisions which clearly do not confer subjective rights onto individuals can nonetheless be invoked by individuals against their state. 34 In order to account for these judgments, some commentators-mainly in the German and Dutch literature-have distinguished between a narrow concept of 'subjective direct effect' and a broader concept of 'objective direct effect' to distinguish between the invocation of an EU provision that confers a subjective right to individuals ('subjective direct effect') and the invocation of an EU provision which does not confer any rights to individuals but which imposes an obligation onto national judicial and administrative authorities to apply this provision ('objective direct effect'). 35 In line with Pescatore's 'infant disease' conception of direct effect, it seems however that most authors see the possibility to invoke any sufficiently precise and unconditional provision of EU law before national courts as further evidence that direct effect is merely a matter of justiciability. 36 The question is whether this assertion is correct in the context of the direct effect of directives. Indeed, after the Court had confirmed the invocability of directives by individuals in proceedings against the state in Van Duyn and subsequent cases, in Marshall it famously observed obiter dictum that the binding nature of directives exists only in relation to the Member States, and that they can, therefore, only create rights for individuals-not obligations. 37 Accordingly, even if a provision of a directive is sufficiently clear and unconditional, an individual cannot rely on it against another individual (the

Squintani and Lindeboom 24
prohibition of horizontal direct effect). 38 An explicit call by Advocate General Lenz to change course and allow for horizontal direct effect of directives was famously rejected by the Court in Faccini Dori. 39 In this case, Ms Faccini Dori invoked the (unimplemented) Doorstep Selling Directive 40 in a dispute against the street seller who had persuaded her to buy a language course and then refused to annul the contract a few days later, as it would have been required to do had Italy implemented the Directive. The Court had no difficulty in rejecting Ms Faccini Dori's position. By reference to Marshall, the ECJ shut the door to horizontal direct effect: 'a Directive cannot of itself impose obligations on an individual'. 41 Likewise, a Member State cannot rely on a sufficiently clear and unconditional provision of a directive against an individual (the prohibition of inverse vertical direct effect). 42 These two prohibitions appear to be at odds with the thesis that, at least in the context of directives, the doctrine of direct effect is merely a question of justiciability. After all, if direct effect can be reduced to justiciability, this would suggest that sufficiently clear and unconditional provisions from (unimplemented) directives should be capable of being invoked even against other individuals. 43  While questions of direct effect also continue to arise in the context of primary law, 45 such questions are mainly associated with directives. Looking at the case law, it appears that the complexity surrounding the direct effect of directives points to unresolved issues in the invocability of EU law. 46 Questions of invocability encompass not only the justiciability of the provision to be invoked, ie the objective capacity of that provision to be applied by courts, but also the question of who is entitled to invoke this provision, against whom it can be invoked, in addition to various procedural limits to the invocability of legal norms.
The complexity of the invocability of directives has been further increased by post-Faccini Dori case law, which has-at least ostensibly-limited the relevance of the apparent clarity of the prohibitions of horizontal and inverse vertical direct effect. 47 The prohibition of the horizontal direct effect of directives has been challenged directly by CIA Security and Unilever Italia, also known as the 'incidental effects' 48 case law, in which the Court allowed a private party to rely on the Notification Directive 49 in a horizontal dispute against another private party. 50 Such 'incidental effects' were seemingly also allowed for a broader range of directives. 51 In every case where the Court was explicitly asked a question about the horizontal and inverse vertical effects of directives, however, it responded by explicitly maintaining the prohibition of horizontal direct effect and inverse vertical direct effect in other cases, 52 causing confusion in the literature. 53 This confusion is especially understandable if chronology is taken into account. Less than two months before the Court's judgment in CIA Security, it had confirmed the prohibition of horizontal direct effect in El Corte Inglés. 54 As a corollary of the 'incidental effects' case law, doubts about the relevance of the limits concerning the prohibition of inverse vertical direct effect also arose in 'triangular situations', where mere adverse repercussions for third parties are perceived as side effects of vertical direct effect. 55  In this context of increasingly complex case law, a number of theories have been developed to try to make sense of the direct effect of directives. In the remainder of this section, we will briefly summarize the most important attempts in this regard, focusing on their implications for the meaning of the doctrine of direct effect more generally. First, we look at theories focusing on the distinction between different factual situations (Section II.A), then on those focusing on the distinction between direct effect and the scope of application of a provision (Section II.B), and finally on those distinguishing between substitutionary and exclusionary effects (Section II.C). Whilst this section will be confined to describing these theories neutrally, we will return to them in Section IV below by scrutinizing their explanatory and normative attractiveness in view of the distinction between 'direct obligations' and 'mere adverse repercussions' which we regard as the central distinction governing the case law, as discussed in Section III.

A. Distinguishing between different situations
Some commentators trying to make sense of the 'incidental effects' case law have suggested to distinguish between different 'situations' in which direct effect plays a role. Consequently, the prohibition of horizontal direct effect of directives may not apply to all situations of invoking a directive in a horizontal dispute.

Squintani and Lindeboom 28
Hilson and Downes aimed at explaining the 'incidental effects' case law by relying on Hohfeld's theory of rights 62 in order to distinguish between two situations: first, an individual uses direct effect as a 'sword' by claiming a subjective right (more precisely using Hohfeld's terminology, a 'right' in the limited sense of a claim 63 ) from a directive which correlates with an obligation (or, to use Hohfeld's terminology, a duty 64 ) on another individual. Secondly, an individual can also invoke an EU provision as a 'shield' to protect himself from obligations under conflicting national law. According to Hilson and Downes, the prohibition of horizontal direct effect would prevent an individual from invoking a provision from an unimplemented directive insofar as the provision contains a right in the Hohfeldian sense. 65 In cases such as CIA Security no rights were invoked, because the Notification Directive simply contains neither rights nor obligations for individuals at all: CIA was not attempting to assert a positive or claim-right derived from the Directive against the defendants, who were also individuals. Rather, they were seeking to enforce a Community law immunity (or negative right) based on the Directive (ie using the Directive as a 'shield'), the correlative of which is that the defendants would be under a disability from relying on the Belgian rules. . . . As a result, CIA were recognised as enjoying a horizontal immunity (or negative right) vis-á-vis the counterclaiming company in this case, despite the orthodox understanding of the absence of horizontal effect for directives. 66 The correlation of this immunity is not an obligation, but a 'disability' on the part of the other individual consisting of the fact that this individual is prevented from relying on national law. Accordingly, a directive can be invoked even in a horizontal dispute if it is used merely as a 'shield' because such a situation cannot properly be characterized as imposing an obligation on the other individual. 67 By contrast, invoking a directive as a 'sword' would be prohibited because invoking the right implies imposing the correlative obligation on another individual. 68  Secondly, in his well-known commentary on the 'incidental effects' case law, Dougan introduced the notion of so-called 'disguised vertical direct effect' in order to conceive of the incidental effects case law as effectively cases involving vertical direct effect. 70 In accordance with this theory, the Court has silently followed only one part of the calls for horizontal direct effect, namely the 'extended estoppel theory': individuals who would have had an obligation if the directive had been implemented correctly, may not benefit from the failure of a Member State to do so. 71 Dougan distinguishes between two types of horizontal disputes. In the first situation, only the Member State has actively breached the directive, while the individual must be regarded as an 'opportunistic passer-by' who tried to benefit from the Member State's failure. For instance, the companies concerned in CIA Security and Unilever had not themselves infringed any EU provision; they merely tried to enforce national law against CIA Security and Unilever. According to the 'disguised vertical direct effect' theory, this situation is comparable to a Member State trying to enforce national law which violates a directive against an individual. 72 Since, in such a situation, the individual can invoke the directive against his obligations under national law, there is no reason why this individual should not also be allowed to invoke the directive if that national law is enforced by another individual rather than the Member State itself. Hence, the position of the private party can be assimilated entirely to that of the defaulting Member State and therefore subjected to the same rule as regards direct effect without affecting anything other than the purely factual interests of that individual (CIA Security; Unilever v Smithkline Beecham; Pafitis). The rationale here seems to be that not only the guilty public authority but also an opportunistic passer-by should be prevented from taking advantage of the Member State's substantive breach of Community law. 73 In the second situation, an individual does not merely try to enforce national law contravening the applicable directive, but has herself or himself actively transgressed the substantive obligations contained in the directive. A typical example is Faccini Dori, where two breaches of the Doorstep Selling Directive can be distinguished: Italy's failure to implement the Doorstep Selling Directive, and the doorstep seller who refused to cancel Ms Faccini Dori's subscription to the language course she had purchased: In this type of case, the interests of the private party are not completely parasitic upon the conduct of the national authorities; enforcement of the Member State's obligation to implement the terms of the directive would necessarily expand upon the private 70 Dougan (n 53). 71 See van Gerven (n 47), 351-3. 72 Dougan (n 53), 606-7. 73 Ibid 607.

Squintani and Lindeboom 30
party's legal responsibilities towards the claimant in respect of its own breach of Community law. 74 B. Distinguishing between the direct effect and the scope of a provision As a second approach to account for the case law on the direct effect of directives, some authors have tried to take an interpretive approach to invocability, focusing on the personal and material scope of provisions. The question of whether a directive has the 'effect' of being applied in a specific procedural context-a horizontal or a vertical dispute-should be distinguished from the question of whether this directive includes in its personal and material scope the actions of private parties. Consequently, the possibility of invoking an unimplemented directive would then not depend on its horizontal or (inverse) vertical direct effect, but on the question of whether the respective provision would impose an obligation onto an individual if it is applied. To the extent that the substantive content of the provision does not impose any obligations onto individuals, the directive could be invoked and applied even in horizontal disputes. 75 What speaks in favour of this approach is that the Court typically does not use these categories either, focusing instead on whether a provision can be invoked in this situation and/or against this (specific) party. 76 Arguably the same applies to the case law on the invocability of the fundamental freedoms. 77 A rather crude formulation of this interpretive approach to the direct effect of directives can be linked to early attempts to explain the 'incidental effects' case law by making a distinction between different types of directives. 78 Some directives only intend to create obligations for the Member States, while others have as their objective the creation of rights and obligations for private parties. 79 74 Ibid. 75 In its various manifestations, this theory has been proposed in the context of the debate on the horizontal direct effect of directives. It is unclear to what extent its proponents would also endorse the same distinction in the context of inverse vertical situations. To this end, see our critique on this theory in section IV.B below. 76 See eg Faccini Dori (n 39), para 20; CIA Security (n 50), para 55. While seeking further support for this hypothesis, we found that a Curia search-which admittedly is a rather crude and all but foolproof method-for 'horizontal direct effect' in grounds of judgment results in only one result,  It is not necessary for this theory, however, to distinguish only between different 'types' of (entire) directives. In its most refined form, the interpretive approach would boil down to reducing the prohibition of horizontal direct effect to a prohibition of giving effect to an unimplemented directive which includes in its scope actions by private parties. 87 By contrast, unimplemented directives which do not cover actions by private parties in their scope could be invoked in horizontal situations without any difficulties, because by definition they do not impose obligations onto individuals. At its core, this theory could be understood as arguing that there is no prohibition of a direct effect of directives, in whatever type of dispute, because all limits to the invocability of justiciable directives could be conceived as interpretive questions about the directive's scope. 88 In practice, therefore, the question of whether a directive may be invoked against another individual can be answered by looking at the ratione personae

Squintani and Lindeboom 32
and ratione materiae of the respective provision. Insofar as the specific provision does not include actions by private parties (ratione materiae), and is not directly aimed at private parties (ratione personae), it can be invoked even in a horizontal dispute. Notwithstanding the horizontal nature of the dispute, the effect of the directive is in reality to obligate the national court to set aside conflicting national law. Insofar as the scope of the provision of the directive does not include actions by individuals, this provision cannot be said to impose any obligations onto another individual, so that the prohibition of horizontal direct effect is not applicable.

C. Distinguishing between direct effect and primacy: the substitutionexclusion dichotomy
The distinction between invocabilité de substitution and invocabilité d'exclusion (hereinafter also referred to as the 'substitution-exclusion dichotomy) originated in the French literature on European administrative law, 89 in particular the work of Denys Simon, 90 and has been actively endorsed by several Advocates General and members of the Court. 91 In its original formulation, invocabilité d'exclusion and invocabilité de substitution are two out of five distinct forms of invocability with increasing intensity of justiciability: l'invocabilité d'interprétation conforme (consistent interpretation); l'invocabilité d'exclusion; l'invocabilité de prevention (the Inter-Environnement Wallonie obligation for Member States to refrain from jeopardizing the future effectiveness of directives even before the end of the transposition deadline); l'invocabilité de réparation (state liability); and l'effet de substitution. 92 Other commentators have mainly focused on the distinction between invocabilité d'exclusion and invocabilité de substitution in order to explain the case law on the direct effect of directives. 93 Whilst there appear to be some nuances 89 See generally, Y Galmot, J-C Bonichot, and G Isaac, Droit communautaire general (Masson, 1994); T Dal Farra, 'L'invocabilité des directives communautaires devant le juge national de la légalité' (1992) 28 RTD eur.i, 631; D Simon, La directive européenne (Dalloz, 1997); Simon (n 1). 90 Simon (n 1). 91  The Normative Impact of Invoking Directives 33 among the various available approaches, in essence the substitution-exclusion dichotomy boils down to distinction between direct effect and an independent application of the primacy of EU law. 94 According to the substitution-exclusion dichotomy, also known as the so-called primacy model, 95 direct effect must be confined to the invocability of the EU law norms themselves to be applied in the dispute at hand, as for example in Van Gend en Loos and Defrenne. 96 In other words, direct effect is necessarily an instance of invocability of substitution, as the EU provision that is invoked substitutes the conflicting national provision. By contrast, the incidental effects cases and triangular conflicts, such as Wells, are merely examples of primacy: a conflict between national and EU law must be solved by disapplying national law in accordance with the basic conflict rule of the primacy of EU law. So conceived, the substitution-exclusion dichotomy is based on the premise that the primacy rule is in itself capable of generating exclusionary effects, with direct effect playing no role, 97 which fits the obligation of national courts 98 and administrative authorities 99 to disapply all national law in conflict with EU law. 100 As Dougan observed, the substitution-exclusion dichotomy assumes a unitary European legal order, in the sense that the primacy of EU law independently can generate exclusionary effects in the national legal orders without the need for direct effect. 101 Secondly, it assumes that invocability of exclusion does not require an EU provision to be sufficiently precise and unconditional, or at least that this threshold does not apply as forcefully as it does to invocability of substitution. 102 In the context of the invocability of directives, the most notable assertion of the substitution-exclusion dichotomy is that the prohibitions of the horizontal and inverse vertical direct effect of directives only apply to invocability of substitution. We will scrutinize this claim in Section IV below, concluding that at any rate it lacks explanatory power towards the case law. 94 See Dougan (n 4); Lenaerts and Corthaut (n 30). 95 Using the terminology of Dougan (n 4). 96 See particularly Lenaerts and Corthaut (n 30). 97 Dougan (n 4), 933. 98 Simmenthal, 106/77, EU:C:1978:49. 99 Fratelli Costanzo (n 56). 100 The emphasis on the obligation of national institutions to exclude, ie disapply, any national provision conflicting with an EU provision that is invoked by an individual, seems to suggest that the substitution-exclusion distinction is broadly similar to the distinction between subjective and objective direct effect. However, this equation would be mistaken since some instances of 'objective direct effect' require substitution of the directive provision. We show this was the case in, eg, Salzburger Flughafen in Section IV below. 101 Dougan (n 4), 943-4. 102 Dougan (n 4), 941-2, criticizing this assumption by pointing out that insofar as sufficient precision and unconditionality are necessary conditions of justiciability, they necessarily apply to any instance of invocation, because if these conditions are not met, by definition it follows that the provision cannot be applied by a court.

D. An impossible task?
It has become almost a commonplace that none of the suggested theories of direct effect of directives succeeds in providing a watertight explanation of the case law. 103 It seems hardly convincing, therefore, to maintain the proposition that the doctrine of direct effect is merely a matter of justiciability in the context of directives, because clearly the criteria of sufficient clarity and unconditionality do not exhaust the question of direct effect.
Moreover, to the extent that direct effect should be better conceived as a matter of the degree of and limits to invocability, no general theory has been able to explain the seemingly arbitrary exceptions to the prohibitions of horizontal and inverse vertical direct effect. 104 Doctrine has given up, it seems: it hardly seems so painful, being forced to acknowledge that the search for a theoretically respectable, watertight descriptive account of the fractured, fumbling case law on the direct effect of Directives is a task fit only for masochists. 105 Nonetheless, we claim to be able to provide a theory of the case law that does have descriptive accuracy and, further, is normatively desirable over its main competitors. In order to introduce this theory-which we refer to as the 'normative impact theory'-in Section III we will first revisit the case law on the direct effect of directives and highlight that the distinction between 'direct obligations' and 'mere adverse repercussions' is the key distinction in both the case law on the prohibition of horizontal direct effect as well as that on inverse vertical direct effect and triangular situations. Then, in Section IV, we will return to the abovementioned theories to show why they fail to describe the case law: they are unable to accurately reflect the direct obligations-mere adverse repercussions distinction. In Section V, we introduce the normative impact theory, and link it to the case law.

III. Mere adverse repercussions and the prohibitions of horizontal and inverse vertical direct effect
From the perspective of the estoppel argument, 106 the prohibition of inverse vertical direct effect can be viewed as a subject-based limit to direct effect, as the estoppel rule justifies why a certain actor may not invoke a directive's norms. 107 Alternatively, when we focus on the effects of invoking a provision of an EU 103 See Dougan (n 4), 963; Prinssen (n 1). 104 See n 5 above. 105 Dougan (n 4), 963. 106 P Pescatore, 'L'effet des directives communautaires'(n 1), 171; Schütze (n 87), 97-8. 107 Cf vertical and 'intermediate horizontal' situations (the latter term is used by Albors-Llorens (n 42) to describe a dispute between two public authorities), where the estoppel rule gives a justification as to why a certain actor may not resist the invocation of norms of a directive by another actor.
The Normative Impact of Invoking Directives 35 directive, the same prohibition can also be justified by reference to the hypothetical consequences of inverse vertical and horizontal direct effect, namely that an individual would be confronted with legal obligations from an instrument which is only binding on the Member States. Various legal principles, including legality and legal certainty, resist such an outcome. Both subject-based and effects-based arguments can be found in the Court's case law. 108 Yet, all judgments emphasize the fact that directives may not impose obligations onto individuals, thus stressing the consequences of allowing a directive to be invoked to the detriment of an individual. This is most clearly visible in the case law on the prohibition of inverse vertical direct effect of directives. The first case in which the Court clarified the existence of a prohibition of inverse vertical direct effect concerns the criminal proceedings in Pretore di Salò. 109 An Italian criminal court brought a prosecution against unknown persons for the pollution of the waters of the river Chiese in Italy. The Italian court leading the preliminary investigations wanted to base its proceedings on Articles 635, 625(7), and 632 of the Italian Criminal Code on the aggravated pollution of waters, diversion of water and interference with the state of premises respectively. 110 However, this action required considering the river Chiese as being a habitat for fish, something which would have been possible only by relying on Directive 78/659/EEC on the quality of fresh waters itself. 111 Italian authorities had allegedly failed to designate the river Chiese as salmonid waters in need of protection under Article 4(1) of the Directive and thus did not adopt a programme of measures to protect the river, as would have been required by Article 5 of the Directive. The national court did not enquire whether the Directive may of itself and independently of the internal law of a Member State could have the effect of determining or aggravating the liability in criminal law of persons acting in contravention of the provisions of that Directive. Yet, the ECJ decided to interpret the preliminary question to include this issue, to which it replied by stating that Council Directive 78/659/EEC cannot, of itself and independently of the implementing legislation adopted by a Member State, have the effect of determining or aggravating the liability in criminal law of persons who act in contravention of the provisions of that Directive, without reference to the estoppel argument. 112 108 Sometimes concurrently, see eg Kolpinghuis (n 42), para. 8 (subject-based arguments) and para. 9 (effects-based arguments). 109

Squintani and Lindeboom 36
This conclusion was reiterated in subsequent case law on criminal proceedings, 113 with Kolpinghuis becoming the landmark case on the prohibition of inverse vertical direct effect. 114 The principles of legality and legal certainty, with their corollaries-most notably, the principle that a provision of the criminal law may not be applied extensively to the detriment of the defendant, and the principle of non-retroactivity of a criminal rule-preclude criminal proceedings in respect of conduct not clearly defined as creating criminal liability by law. 115 The only exception here is a situation in which criminal liabilities are based on more stringent protective measures, ie gold-plating and green-plating, as this is essentially national law subject to EU general principles and fundamental rights only when restricting the achievement of EU goals. 116 Such a situation occurred in Fornasar, 117 where an Italian court asked several questions about the legal status of diphenylmethanedi-isocyanate (MDI) under Council Directive 91/ 689/EEC on hazardous waste. 118 While this substance is not indicated as hazardous under the then applicable Community rules, MDI is extremely dangerous to human health according to experts, and it could be included in one of the annexes to the Directive due to its similarities with the substances indicated therein. 119 Mr Fornasar and the other private persons facing criminal proceedings explicitly relied on the prohibition of inverse vertical direct effect to argue that the case was not admissible. 120 The Court dismissed this plea, as it was for the national court to decide whether to ask preliminary questions and there was no evidence that the answer to the questions did not bear on a real situation or on the subject matter of the case in the main proceedings. 121 The Court then concluded that the Directive does not prevent the Member States, including their courts for matters within their jurisdiction, from classifying as hazardous  The Normative Impact of Invoking Directives 37 waste anything not featuring on the list of hazardous waste laid down by Community law, and thus from adopting more stringent protective measures in order to prohibit the abandonment, dumping or uncontrolled disposal of such waste. 122 Despite the relevance of this effects-based argument in the field of criminal law, it would be mistaken to consider it relevant only in the context of criminal proceedings. 123 The prohibition of inverse vertical direct effect also applies in the context of administrative law proceedings. In Portgás, 124 the manager of the Operational Programme North, which is part of the Portuguese Ministry of Agriculture, the Sea, the Environment and Town and Country Planning, wanted to rely on Article 4(1) of Council Directive 93/38/EEC 125 to recover the financial assistance which had been granted to Portgás. The ECJ replied that directives cannot impose obligation upon individuals and that if Portgás were to be qualified as a private person, the Portuguese state would be precluded from relying on the Directive against Portgás, 126 something which was for the national court to ascertain.
Finally, the prohibition of inverse vertical direct effect is also relevant in private law proceedings. In Accardo, 127 the ECJ considered whether the Turin Municipality could rely on Council Directive 93/104/EC 128 on working time, against Mr Accardo and other municipal police officers. Mr Accardo and his colleagues had claimed compensation for harm allegedly suffered in the period 1998-2007 as a result of the Municipality's failure to comply with requirements on weekly rest periods. The ECJ explicitly denied the Municipality the possibility of relying on the Directive against Mr Accardo and his colleagues. 129 Three red lines emerge from the analysis above. First, the estoppel argument has sometimes been used by the ECJ as an autonomous argument to justify the prohibition of inverse vertical direct effect. In addition, however, the ECJ has indicated that the principles of legality, legal certainty, and more implicitly, proportionality 130 stand in the way of applying directives directly against individuals, 131 overall stressing the importance of considering the consequences of 122 Ibid para. 51. 123 cf Lenaerts and Corthaut (n 30), 303; see further section IV.C below. 124 Portgás (n 6). 125 Council Directive 93/38/EEC coordinating the procurement procedures of entities operating in the water, energy, transport and telecommunications sectors [1993] OJ L199/84. 126 Portgás (n 6) para 25. 127 Accardo and Others C-227/09, EU:C:2010:624. 128 Council Directive 93/104/EC concerning certain aspects of the organization of working time [1993] OJ L307/18. This provision establishes an exception to certain provisions of the directive. 129 Accardo (n 127), paras 46-47. 130 The principle of proportionality is relevant in that Art. 5(4) TEU obligates the EU to use the least onerous legal instrument. It is widely accepted that a directive allows the Member States more room for discretion than a regulation does, with only the latter having direct applicability vis-à-vis individuals. See also Faccini Dori C-91/92, EU:C:1994:292, paras 22-24. 131 See also Lenaerts and Corthaut (n 30).
Squintani and Lindeboom 38 direct effect. Indeed, in some cases the prohibition of inverse vertical direct effect was only justified by reference to effects-based arguments. 132 In this regard, the Court stressed, third, that the prohibition of inverse vertical direct effect is triggered by the invocation of a norm which creates obligations. Invoking a norm which only results in factual repercussions-as opposed to obligations-does not trigger the prohibition of inverse vertical direct effect. This contrast is shown most explicitly by the Wells case. 133 In this case the Court allowed Mrs Wells to rely directly on a provision of the EIA [Environmental Impact Assessment] Directive 134 in challenging a decision of a public authority granting development consent to an undertaking for the exploitation of a quarry adjacent to Mrs Wells's house. According to the Court, the annulment of the consent is a 'mere adverse repercussion on the rights of [the quarry owner]'. Even if such repercussions are certain, they 'do not justify preventing an individual from relying on the provisions of a directive against the Member State concerned'. 135 In this regard, the ECJ emphasized that triangular situations are only possible when the vertical direct effect of the Directive leads to mere adverse repercussions-and not a legal obligation-for a third private party. 136 This showed that the ECJ explicitly aimed to maintain the prohibition of inverse vertical direct effect highlighted in Kolpinghuis. 137 While triangular situations are common mainly in environmental law cases, the doctrine extends equally to other areas of administrative law, as can be seen for example in Fratelli Costanzo (public procurement), 138 and Arcor (telecom regulation). 139 The obligations versus mere adverse repercussions dichotomy was maintained in Salzburger Flughafen, regardless of the fact that, as opposed to Wells, in this case it was a public authority that relied on the EIA Directive against another public authority. The facts of this case were roughly as follows. In 2004 Salzburger Flughafen GmbH submitted two requests for development consent for the expansion of the airport of Salzburg in Austria, to the relevant competent authority (the Amt der Salzburger Landesregierung). An EIA for these two projects was not required under national law because the EIA Directive had been transposed into national law to include a threshold which was not trespassed by the project ad quo. The Umweltsenat, however, was of the opinion that national law violated the Directive, and applied the relevant provisions of the Directive directly by establishing that the development consent was conditional on an EIA 132 Pretore di Salo (n 42), para. 20. See also, Berlusconi (n 42), paras 73-74. 133  The Normative Impact of Invoking Directives 39 being produced. This decision subsequently led to legal proceedings before the Austrian Supreme Administrative Court (Verwaltungsgerichtshof ), which referred a number of preliminary questions to the ECJ as to the proper interpretation of the Directive. As regards the question of whether the Directive could be applied directly, the ECJ concluded that Salzburger Flughafen, as the user of the land in question, must also bear the consequences of such a decision. 140 This conclusion shows that the Court's focus on the obligations versus mere adverse repercussions dichotomy is agnostic to the question of who is invoking the directive, further decreasing the relevance of the estoppel rationale. The same is true for the case law on the prohibition of horizontal direct effect. In this context, while the Court does not use the terminology of direct obligations and mere adverse repercussions, the main argument used by the Court to limit the invocability of directives in horizontal situations is intrinsically linked to the obligations-mere adverse repercussions dichotomy, as a closer look at the case law reveals.
In judgments affirming the prohibition of horizontal direct effect, the Court invariably emphasized that directives cannot be invoked against individuals because they cannot of itself impose obligations on an individual. 141 Accordingly, unlike in the case law on inverse vertical direct effect, the question of who is invoking a directive plays no role in the case law on horizontal direct effect, which is based entirely on the consequences which an individual must suffer-in line with the obligations versus mere adverse repercussions dichotomy.
The case law on the so-called incidental effects provides further confirmation of the Court's focus on this dichotomy in horizontal disputes, with Unilever Italia as the main exponent. As we noted in Section II above, CIA Security, 142 Pafitis, 143 and Ruiz Bernáldez 144 had led commentators to question the relevance of the prohibition of horizontal direct effect. 145 When the Court was asked to clarify the relevance of CIA Security for the prohibition of horizontal direct effect of directives, it responded by stating that 'Directive 83/189 does not in any way define the substantive scope of the legal rule on the basis of which the national court must decide the case before it. It creates neither rights nor obligations for individuals'. 146 In  145 See eg n 53 and n 56. We discuss the facts of these cases further in Sections V and VI. 146 Unilever (n 50), paras 50-51.

Squintani and Lindeboom 40
agency El Corte Inglés because that would impose an obligation on the latter. 147 Similarly, a few years after the incidental effects cases in its Pfeiffer judgment, which concerned the Working Time Directive, 148 the Court concluded that reliance of the Directive would impose an obligation upon individuals that was not permissible. 149 A conjunctive look at Unilever and Pfeiffer shows that, once again, the obligations versus mere adverse repercussions dichotomy is used as a cornerstone for the invocability of direct effect in the context of directives. Implicitly, in CIA Security and Unilever Italia, the Court considered the negative effects of invoking the Notification Directive for the other individuals as mere adverse repercussions. In Pfeiffer it explicitly observed that invoking the Working Time Directive would entail direct obligations for another private party.
This distinction is most recently supported and clarified by Smith, in which the Court revisited its jurisprudence. In this case, Mr Smith became seriously injured in a car accident while he was sitting in the rear of a van owned by Mr Meade. This rear part of the van was not designed and constructed with seating accommodation. Consequently, Meade's motor insurer FBD refused to provide him an indemnity on the basis of its insurance policy which did not cover personal injuries to persons sitting in part of a vehicle not designed and constructed with seating accommodation. This policy was in conformity with Section 65(1)(a)(i) of the Irish Road Traffic Act. According to Article 1 of Directive 90/232/EEC, however, insurance against civil liability in respect of the use of motor vehicles 'shall cover liability for personal injuries to all passengers, other than the driver, arising out of the use of a vehicle'. 150 In response to the question of whether FBD's insurance policy ought to be disapplied for violating Article 1 of the Directive, the Court answered in the negative. It confirmed that 'a directive cannot of itself impose obligations on an individual and cannot therefore be relied upon as such against an individual', 151 and distinguished CIA Security and Unilever Italia from Faccini Dori and Pfeiffer by emphasizing that Directive 83/189/EEC created neither rights nor obligations for individuals, did not determine the substantive content of the legal rule on the basis of which the national court had to decide the case before it, meaning that the case-law to the effect that a directive that has not been 147 El Corte Inglés (n 52). 148 Council Directive 93/104/EC of 23 November 1993 concerning certain aspects of the organization of working time [1993]  Linking the notion of imposing direct obligations on individuals to the determination of the substantive content of the rule by the national court to the detriment of a private party, the answer to the question what demarcates 'direct obligations' from 'mere adverse repercussions' is related to the question of whether invoking a directive affects the substantive content of the rule which determines the dispute at hand. The next section looks at how the doctrine described in Section II above copes with this task.

IV. Existing theories of the direct effect of directives and the distinction between direct obligations and mere adverse repercussions
Following the descriptive overview of the main theories that have tried to explain the Court's case law on the direct effect of directives in Section II above, we will now discuss the viability of these theories in the same order in light of their ability to cope with the distinction between imposing direct obligations and mere adverse repercussions, highlighted in Section III above. Although some of these theories have been criticized extensively in earlier literaturesometimes by their respective authors themselves 154 -we submit that this analysis is useful for the purpose of this article. First, we will ground our critique of all theories in their failure to account for the distinction between direct obligations and mere adverse repercussions. Secondly, this section will not only serve as a 'negative' analysis of existing theories, but will also pave the way for the introduction of the normative impact theory of direct effect of directives. An analysis of existing theories based on the distinction between direct obligations and mere adverse repercussions will help to understand why the normative impact theory does succeed in explaining the case law.
In the subsequent analysis, we will pay somewhat more attention to the descriptive and normative force of the substitution-exclusion dichotomy, 152 Smith (n 6), para. 53 (emphasis added). 153 On the relevance of the functioning of the preliminary reference procedure for the case law on direct effect, see further Section VI below. On the difference between Smith and Ruiz Bernáldez see further Section V.B.i. below. 154 eg Hilson and Downes (n 53) and Dougan (n 53). because it is arguably the most well-known theory on the direct effect of directives, having gained considerable support in the literature, and that of several Advocates General. 155

A. Theories based on different factual situations
As noted in Section II.A, Hilson and Downes' analysis of the horizontal direct effect case law in light of Hohfeldian relations focuses on the distinction between the two correlatives right-obligation and immunity-disability. In this respect, it seems a suitable candidate for explaining the case law to the extent that the prohibitions of horizontal and inverse vertical direct effect of directives are rooted in the notion of directly imposing an obligation onto an individual. However, as Hilson and Downes readily admit, their approach fails to explain the Court's jurisprudence. 156 In Pafitis, for example, the old shareholders' invocation of the directive should in Hohfeldian terms be deemed a 'right' to vote on any capital increase, with the correlative obligation for the bank to annul the decision. Similarly, Ruiz Bernáldez involved the invocation of a right from the respective directive against another individual. 157 Faccini Dori, on the other hand, concerned a situation where Ms Faccini Dori invoked the Doorstep Selling Directives as an 'immunity' rather than a right, entailing a 'disability' on the part of the counter-party rather than an 'obligation'. Hence, contrary to the Court's ruling, the Hohfeldian framework would not predict that Faccini Dori and CIA Security would result in the same outcome. 158 In turn, Hilson and Downes suggest that in a horizontal dispute containing a 'public law element' individuals have a right to invoke directives either as a 'shield' (ie an immunity correlating in a disability, as in CIA Security) or a 'sword' (ie a claim-right correlating in an obligation for the other individual, as in Pafitis and Ruiz Bernáldez). This conclusion is incompatible with the Hohfeldian framework, and is not endorsed by the authors. 159 The explanatory weaknesses of this theory have been further confirmed by Pfeiffer, where the applicable directive was also invoked as an immunity rather than a right, so that the Hohfeldian framework arguably would predict an allowance of horizontal direct effect. In conclusion, Hilson and Downes' theory, although it might have been a viable option for the Court to pursue, rests on a notion of 'obligations' 155 Though the distinction between different 'types' of directives, as described in Section II.B above, is partially rooted in the Opinions of AG Elmer (in CIA Security) and AG Jacobs (in Unilever Italia), neither of these Opinions unequivocally aimed at creating or endorsing a doctrinal theory. By contrast, the substitution-exclusion theory has been clearly supported by AG Léger in Linster (n 91); AG Ruiz-Jarabo Colomer in Pfeiffer (n 91); and AG Kokott in Berlusconi (n 91). See also Lenaerts and Corthaut (n 30). 156 Hilson and Downes (n 53), 126. 157 Ibid 125-6. 158 Ibid 125. 159 Hilson and Downes (n 53), 127. that is clearly not supported by the Court. 160 It might be added more generally that the Hohfeldian framework would clearly fail as an explanatory theory of direct effect since, as we observed above and Hilson and Downes also recognized, direct effect is not restricted to the invocation of rights. 161 We can thus disband this theory, both as a theory of direct effect in general and, specifically, as an explanatory mechanism for the Court's distinction between 'direct obligations' and 'mere adverse repercussions'.
As a second theory distinguishing between different types of factual situations in which direct effect is relied on, Dougan's so-called 'disguised vertical direct effect' theory conceives the incidental effects case law as cases involving vertical direct effect. 162 Dougan's theory can be restated as follows: a provision from an unimplemented directive can be invoked against an individual insofar as that individual has not 'actively breached' the provision but is merely trying to benefit opportunistically from the state's failure to implement the directive. In light of the distinction between direct obligations and mere adverse repercussions, such a situation of 'disguised vertical direct effect' should not be conceived as imposing an obligation onto the other individual, for she or he has not breached any provision from the directive.
Thus rephrased, it becomes clear that in any case Berlusconi seems to remain an anomaly in the case law. In that case, after all, Mr Berlusconi should be qualified as merely an opportunistic 'passer-by' who tried to benefit from new Italian legislation which breached Italy's obligations under the directive. This conclusion cannot be altered by the fact that Berlusconi de facto played a political role in introducing it.
Although his theory seems to fit most of the case law apart from Berlusconi, Dougan already noted himself that the perverse consequence of this approach is that an individual who has actively infringed a directive is protected by the prohibition of horizontal direct effect, while an individual who has not actively breached the directive will be exposed to the directive's 'disguised vertical' direct effect despite being less culpable. 163 Moreover, the theory is premised on the assumption that some directives, from a material viewpoint, can be breached by individuals, the denial of which is the very basis of the prohibitions of inverse vertical and horizontal direct effect. Accordingly, Dougan's explanation is deeply at odds with the theoretical assumptions of the Marshall and Kolpinghuis case law. 160 From a Hohfeldian perspective, this might be a loss for the Court, not theory. One of Hohfeld's objectives indeed to expose 'rights-talk' by legal participants is often conceptually confused, as the manner in which legal participants usually talk about rights 'are a peril both to clear thought and to lucid expression'. WN

Squintani and Lindeboom 44
These general criticisms gain further force if the distinction between direct obligations and mere adverse repercussions is put in the spotlight. The question of whether invoking an unimplemented directive against an individual amounts to imposing an obligation onto the latter cannot solely be based on the question of whether that individual has 'breached' the directive. By emphasizing the effects-based rationale of the prohibitions of horizontal and inverse vertical direct effect (see Section III above), the case law emphasizes the question of whether invoking a directive results in imposing a direct obligation on an individual, remaining indifferent to the question of whether that obligation is caused by that individual having breached the directive herself or himself.

B. Theories based on different types of directives and the scope of their provisions
Theories purporting to draw a distinction between directives which only aim at creating obligations for Member States, and directives which aim at directly regulating private relationships, thus creating rights and obligations for individuals, focus on the question of whether an invoked provision can be said to impose a direct obligation onto an individual. In this respect, it also seems suitable to explain the dichotomy between direct obligations and mere adverse repercussions.
Yet, as Prinssen points out, if we focus on the 'type' of directive and/or its 'objective', Pafitis already entails complications in this respect, as the relevant Directive concerns the alteration of company capital, and thus clearly intends to regulate the rights and obligations of shareholders and executives. 164 Nonetheless, the Court accepted direct effect contrary to the theory's prediction. We might add that Ruiz Bernáldez fails to fit into this approach as well, as the applicable Directives relate to insurance in respect of civil liabilities and, hence, regulate private relationships. 165 More generally, there are provisions of directives that are exclusively aimed at creating obligations for Member States' but which are nevertheless clearly capable of resulting in the imposition of direct obligations upon individuals if they were to be invoked before a national court. This is for example the case for the vast majority of directives adopted in the field of environmental law. They typically create obligations for the Member States or their national authorities 164 Prinssen (n 1), 162-3. 165 168 The growing tendency of relying on wrongly implemented EU environmental law provisions in torts based cases, as occurred in the Urgenda case, 169 even in horizontal situations, 170 suggests that the same conclusion could be reached in cases started by private parties. In turn, the prohibitions of inverse vertical direct effect and horizontal direct effect of directives would be superfluous in the context of EU environmental law, a conclusion which has no support in the case law.
Moreover, this theory assumes that there is a distinction between directives which, in substance, do create obligations for individuals, and directives which do not, and that this characteristic has repercussions for the question of whether a certain directive can have direct effect. However, the denial of directives being able to impose direct obligations onto individuals lies precisely at the core of the case law which doctrine ought to explain. In sum, this theory fails to explain how distinguishing between different types of directives can make sense of the distinction between imposing direct obligations onto individuals and causing mere adverse repercussions.
Making a distinction between the direct effect of a provision and the interpretation of its personal and material scope, as Schütze suggests, 171 causes the same explanatory and normative problems. Such an approach may serve to explain the outcome of the incidental effects case law, where it was emphasized that the Notification Directive did not contain any obligations for individuals and, accordingly, could be invoked in horizontal disputes. 172 Similarly, cases focusing on the EIA Directive, such as Wells and Salzburger Flughafen, could be interpreted as implying that direct effect was allowed because the respective provisions of the EIA Directive only contain obligations for Member States, excluding individuals from their personal scope. 173 However, in contrast to situations of horizontal direct effect, this approach is irreconcilable with the case law on inverse vertical situations of an administrative or criminal nature, since directives in these areas typically contain ample provisions merely directed towards Member State actions. The Berlusconi case serves as a useful example. 174 By definition, the personal scope of the provisions of the Company Directive that require Member States to provide for appropriate penalties against money laudering are confined to Member State actions. Accordingly, these provisions can never directly cover 'private party actions'. Yet, in Berlusconi direct effect was denied because the invocability of the unimplemented directive clearly caused the imposition of an obligation onto Mr Berlusconi. If the prohibition of inverse vertical direct effect of directives would not to apply to provisions excluding private party actions from their personal and/or material scope, Member States could freely apply any provisions such as those requiring Member States to take appropriate, proportionate, and deterrent sanctions from unimplemented directives against individuals.
The same conclusion applies to directives in the field of EU environmental law, where, as we already noted above, the overwhelming majority of provisions cover only Member State actions. To say these provisions could be invoked in horizontal disputes, in vertical disputes in which an individual seeks to force the Member State to take enforcement action against another individual, or in inverse vertical cases, would be incompatible with the case law.
A counterargument against this criticism could of course be that the prohibition of inverse vertical direct effect has an independent rationale, ie the estoppel rule, which prevents a Member State from relying on any directive against an individual. However, the estoppel rule has been criticized extensively for being unsuitable as a basis for direct effect in general, as direct effect would be merely a 'side effect' or 'corollary' of the failure of the Member State concerned. 175 Moreover, in the context of the prohibition of inverse vertical direct effect, the Court has consequently emphasized the effect-based rationale of the prohibition of inverse vertical direct effect. 176 In Arcaro, while the Court refers to the estoppel argument in the context of the prohibition of inverse vertical direct effect of directives, the prohibition itself is clearly grounded in the imperative of avoiding the imposition of direct obligations upon individuals: It follows that a directive may not by itself create obligations for an individual and that a provision of a directive may not therefore be relied upon as such against such a person. The Court has stated that this case-law seeks to prevent a Member State from taking advantage of its own failure to comply with Community law. In that same line of authority the Court has also ruled that a directive cannot, of itself and independently of a national law adopted by a Member State for its implementation, have the effect of determining or aggravating the liability in criminal law of persons who act in contravention of the provisions of that directive. 177 The relevance of the estoppel argument as a foundation of the prohibition of inverse vertical direct effect is also denied in Salzburger Flughafen, where the Court allowed an administrative entity to rely on a directive against another administrative entity of the same state. 178 This conclusion confirms that the estoppel argument itself is no longer relevant for the question of whether an emanation of the state can rely on an unimplemented directive, leaving only the question of whether invoking a directive would lead to direct obligations for an individual. 179 In sum, after stating that it is relevant to look at whether a directive aims to create obligations for the Member States, or for private parties, this approach fails to provide convincing and useful guidance about how to understand this differentiation, whether or not this theory is framed as distinguishing 'types of directives' or as looking at the personal and material scope of the specific provision in question. In both cases, the theory fails to appreciate that the 175 Prechal (n 13), 223-6, with further references. 176 Pretore di Salo (n 42), para. 20; Berlusconi (n 42), paras 73-74. 177 Arcaro (n 42), paras 36-7 (emphasis added and references to earlier case law omitted). 178 Salzburger Flughafen (n 7). 179 Salzburger Flughafen (n 7), paras 46-47: 'In the case which gave rise to the Wells judgment, the Court held, firstly, that it had to be recognised that it is possible for an individual to rely on the provisions of Directive 85/337 and, secondly, that the owners of the land at issue had to bear the consequences of the belated performance of the obligations of the Member State concerned which follow from that directive. Thus, in the main proceedings [where the EIA Directive was relied upon by an administrative authority], in the event that a decision finds that an environmental study is necessary, Salzburger Flughafen, as the user of the land in question, must also bear the consequences of such a decision'. See also Squintani and Vedder (n 140).
prohibitions of horizontal direct effect and inverse vertical direct effect are effects-based in that they serve to avoid imposing obligations onto individuals as a consequence of invoking a directive. Invoking a directive which only covers Member State actions as its personal and material scope is nevertheless liable to create direct obligations for other individuals, making the theory under-inclusive with a view to the rationale of the Marshall case law.

C. The substitution-exclusion theory
Also the well-known substitution-exclusion theory fails to cope with the distinction between direct obligations and mere adverse repercussions. Both general assumptions of the substitution-exclusion theory-the possibility that primacy can generate exclusionary effects independently of direct effect-and the inapplicability of the criteria of direct effect to invocability of exclusion 180 are already in themselves deeply contested. 181 In this section, we will specifically focus on the capability of the substitution-exclusion theory to account for the distinction between direct obligations and mere adverse repercussions in the case law.
This theory's explanatory attractiveness is mainly rooted in the fact that both the CIA Security case law and Wells suggested that where a directive norm is invoked to set aside a conflicting national law norm, individuals must bear the negative consequences of any remaining national law norm or norms which become applicable because of the disapplication of the norm in conflict with the directive. 182 Thus, it would appear that in situations of invocability of exclusion, it is not the directive invoked that imposes an obligation onto an individual. Any such obligations are imposed by the national law remaining in place after exclusion.
However, in Berlusconi, 183 handed down about a year after Wells, the ECJ confirmed that it does not matter whether the norm imposed upon the 180 See Section II.C above. 181 The second assumption is problematic in light of the link between the criteria for direct effect and justiciability, see Section II.C above. As for the first assumption, Dougan observed that the primacy model presupposes a unitary European legal order, in the sense that the primacy of EU law independently can generate exclusionary effects in the national legal orders without the need for direct effect. Criticizing the primacy model of the substitution-exclusion dichotomy, he rightly argued instead that applying primacy necessarily implies direct effect (Dougan (n 4), 942-3) In other words, while the proponents of the substitution-exclusion dichotomy posit a narrow conception of direct effect involving only the application of a sufficiently precise and unconditional norm of EU law to solve a dispute before a national court, this cannot negate the fact that EU law norms must first penetrate the national legal orders before they can have any effect at all. This is a fundamental aspect of direct effect originally conceived by the Court as a corollary to the autonomy of the EU legal order (to this extent, see Van Gend en Loos, 26/62, EU:C:1963:1). We can of course change labels, but this does not resolve the underlying conceptual problem that primacy presupposes direct effect more broadly defined. 182 Ibid. One could infer this from the phrase 'another obligation falling, pursuant to that directive, on a third party' (emphasis added). 183 Berlusconi (n 42).
The Normative Impact of Invoking Directives 49 individual as a result of inverse vertical direct effect is a national rather than an EU norm. In both cases, such an imposition would be contrary to the prohibition of inverse vertical direct effect. By contrast, as Advocate General Kokott argued in her Opinion to this case, 184 according to the substitution-exclusion theory the Directive could be invoked because it would only have an exclusionary effect. 185 Although the Court's explicit reliance on the prohibition of inverse vertical direct effect in Berlusconi strongly suggests that it rejects the substitution-exclusion dichotomy relied upon by Advocate General Kokott, Lenaerts and Corthaut emphasize the fact that the obligation for Mr Berlusconi was of a criminal nature in order to support the substitution-exclusion distinction outside criminal law situations: The problem is not that Community law does not know a technique to set aside contrary provisions, even between individuals, as will appear below, but that in the context of a criminal proceeding the effect of the principle of primacy needs to yield to the legality principle of Art.7 ECHR. 186 However, primacy is interpreted in light of the principles of legality and legal certainty outside criminal law as well. As Article 4(2) TEU shows, the EU shall respect national identities, including respect for the fundamental constitutional structures of the Member States. 187 The case law of the ECJ in private law cases, such as in Pfeiffer, and administrative law cases, such as in Portgás, shows that the application of primacy is equally influenced by the principles of legality and legal certainty in private and administrative proceedings. In Pfeiffer, for example, contrary to Advocate General Ruiz-Jarabo Colomer's endorsement of the substitution-exclusion distinction, 188 the Court quickly moved on to consistent interpretation after having established that directives cannot apply in proceedings between private parties even if they meet the conditions for direct effect. 189 184 AG Kokott in Berlusconi (n 91), para. 150. 185 It should be noted that seven months before Berlusconi, the Court had followed Advocate General Kokott in Criminal proceedings against Antonio Niselli C-457/02, EU:C:2004:707, in which the exclusionary effects of invocability were very similar to those in Berlusconi, leading to a conclusion which strongly suggested an endorsement of the substitution-exclusion dichotomy. However, in contrast to the Niselli judgment, which was rendered by the Second Chamber, Berlusconi was a Grand Chamber case. Hence, it can be safely stated that insofar as Niselli is an anomaly in the case law that has been clearly overruled by Berlusconi. See Dougan (n 4), 954-5. 186 Lenaerts and Corthaut (n 30), 303. 187 Contrary to the view of Lenaerts and Corthaut (n 30), 303, it would be less accurate to say that primacy 'yields' to the principle of legality or other legal principles. Rather, the primacy (or supremacy) of EU law is shaped by these and other general principles as well as Art. 4(2) TEU through a process of adopting principles from other legal systems into the EU legal system.

Squintani and Lindeboom 50
Explicit suggestions by Advocates General to draw a distinction between exclusion and substitution met the same fate in two other cases. 190 These cases show that the substitution-exclusion dichotomy focuses too much on the origin of the obligation, and too little on whether invoking a directive affects the substantive content of the rule which determines the dispute at hand. We can therefore conclude that the substitution-exclusion dichotomy is not viable as a descriptive theory for the case law on direct effect.
The lack of descriptive accuracy does not necessarily mean that the theory is not normatively desirable. By normative desirability, we refer to the ability of a theory to produce outcomes which make sense in light of the distinction between direct obligations and mere adverse repercussions. Accordingly, a theory ought to respect the balance between ensuring the effectiveness of EU law and the principles of legality and legal certainty. As to the distinction between direct obligations and mere adverse repercussions, the substitution-exclusion theory presupposes that this distinction corresponds to the difference between applying a norm from an EU directive, and applying a national norm subsequent to an exclusionary effect.
Hence, the substitution-exclusion dichotomy presents a formalistic interpretation of the prohibition of inverse vertical direct effect, in which the effect of invocability is not relevant, but solely the normative source of a direct obligation. 191 It is readily admitted that 'formalism' runs the risk of being an empty charge, especially in a discipline that relies so much on form as the legal one. 192 What we mean by 'formalistic', in this case, is that the substitution-exclusion theory has an inherent capability to produce false negative or positive outcomes, in light of the aim of preventing direct obligations for individuals as a result of invoking a directive.
A false negative outcome would be a situation in which, despite the prohibition of invoking directives against individuals, an individual is confronted with direct obligations that are a result of invoking a directive due to its purely exclusionary effect. For example, in Berlusconi, the imposition of a criminal section on Berlusconi was the direct corollary of invoking the Directive in order to exclude the application of the national act de-penalizing the conduct. Nevertheless, Advocate General Kokott argued that direct effect should have been allowed because it only led to exclusionary effects.
A false positive outcome would be a situation in which direct effect is not allowed because it entails substitution effects, notwithstanding that the consequences of invoking the directive should be qualified as mere adverse repercussions. We can illustrate such an outcome with Salzburger Flughafen. As noted in Section III above, this case concerned a request to annul a permit granted to an undertaking operating the airport of Salzburg. Similarly to the Wells case, the authorization had been granted without performing an EIA. An EIA for this project was not required under national law because the EIA Directive had been transposed into national law by stating that certain projects are subjected to an EIA only when they trespass a given threshold, 193 which was not the case for the project at stake. Accordingly, this infringement of the Directive could not be remedied by means of national law. In Wells, by contrast, the national obligation to perform an EIA was potentially capable of covering projects such as that at stake in that case, at least to the knowledge of the Court. 194 Hence, while in Salzburger Flughafen only substitution could have created an obligation to perform an EIA, in Wells an exclusionary effect would have sufficed to remedy the deficiency of national law in light of the Directive. Consequently, pursuant to the substitution-exclusion theory, direct effect should not have been allowed in Salzburger Flughafen, while in Wells it should have. The outcome of both cases was the same, however, and this seems normatively sound because direct effect only produced mere adverse repercussions to the opposite private party in both Wells and Salzburger Flughafen.
The inability of the substitution-exclusion theory to cope with the difference between direct obligations and mere adverse repercussions also entails that this theory fails to provide a consistent view on the balance between ensuring the effectiveness of EU law and the principles of legality and legal certainty. Lenaerts and Corthaut, for example, argue that direct effect is 'so powerful that . . . the Court's case law, most notably in respect of Directives, . . . put limits as to whether or when it can be used '. 195 In other words, the legal certainty of individuals must trump effectiveness in cases like Berlusconi and Pfeiffer. This suddenly changes in the context of primacy, the application of which requires no attention for the adverse repercussions for third-party individuals. 196 The primacy model purports to explain the incidental effects case law while maintaining the prohibitions of horizontal and inverse vertical direct effect by simply giving it a different label. 197 This is particularly problematic insofar as the question of Having concluded that all existing theories trying to explain the case law on direct effect of directives fail to make sense of the case law's distinction between direct obligations and mere adverse repercussions, in the next section we suggest to shift perspectives when analysing the limits to the direct effect of directives by looking differently at the question of what is meant by imposing an obligation onto an individual. In this regard, we argue that in determining whether a directive imposes an obligation onto an individual, it is not important whether the norm applied by the national court to solve the dispute is one of national law or of a directive itself. Instead, we focus on the impact of invoking a directive on the norms governing a conflict before a national court or, as shown by Fratelli Costanzo, 199 a public authority. The distinction between the imposition of an obligation contrary to the Marshall rule and mere adverse repercussions for third parties can subsequently be made based on the question of whether invoking a directive modifies the norms which directly apply to the dispute at hand.

V. The normative impact of invoking directives
In this section, we provide a theoretical framework which focuses on the distinction between direct obligations and mere adverse repercussions, and which is capable of explaining all cases involving questions of horizontal or inverse vertical direct effect. This normative impact theory focuses on the impact of invoking an EU directive on the norms directly governing the dispute at hand. We will first introduce the normative impact theory as a doctrinal model based on the direct obligations versus mere adverse repercussions dichotomy (Section V.A). Subsequently, we show how this theory fits the Court's case law (Section V.B), by discussing first the incidental effects case law (Section V.B.i), and then the case law affirming the prohibitions of horizontal and inverse vertical direct effect (Section V.B.ii). 198  A. The distinction between obligations and mere adverse repercussions As mentioned above, in triangular situations direct effect is not possible when it is 'directly linked to the performance of another obligation falling, pursuant to that directive, on a third party'. Rather than focusing on the part of this formulation stating 'pursuant to the directive', as the substitution-exclusion appears to do, we suggest focusing mainly on the part stating 'directly linked to the performance of another obligation'-ie on the concept of 'direct obligation'.
In light of the cases in which the prohibition of inverse direct effect was applied, this formula can be interpreted as meaning that a third party should not be ordered to do or not do something as a direct consequence of direct effect. In the cases in which the Court refused to allow direct effect, had direct effect been allowed the resulting change in the directly applicable norms would have required the national court to order an individual to perform an obligation: the criminalization of a particular kind of conduct following criminal prosecution, 200 the repayment of financial assistance, 201 or the duty to apply a 48hour work week maximum. 202 This is evidently unacceptable under EU law, as it stands.
Yet, in triangular situations such as in Wells, the end of the proceedings does not lead to the imposition of an obligation upon an individual. This is because the action in Wells is an action for annulment against a development consent decision. In this type of action, the success of an individual's claim such as that of Mrs Wells leads to the annulment or suspension of the development consent under dispute. 203 Of course, the obligation to annul or suspend the development consent automatically entails some negative repercussions for another private party, such as the quarry owner in Wells. These repercussions are, however, only factual consequences of the change which occurred in the legal relationship between Mrs Wells and the public authority. 204 Indeed, in Wells we can distinguish between two different factual situations, to each of which applies a separate set of norms. First, there is the factual situation of the quarry owner. He wishes to perform an action-the exploitation of a quarry-and he needs to comply with the norm in UK law which requires him to request and obtain development consent. Secondly, we have the factual situation of the public authority. It has to perform an action-the assessment of the request-and it has to respect the norm in UK law setting out 200 See Pretore di Salò (n 42), Arcaro (n 42), Kolpinghuis (n 42), and Berlusconi (n 42). 201 See Portgás (n 6). 202 See Accardo (n 127), and Pfeiffer (n 10). 203 Wells (n 56), para. 65. 204 Ibid para. 58: 'The fact that mining operations must be halted to await the results of the assessment is admittedly the consequence of the belated performance of that State's obligations. Such a consequence cannot, however, as the United Kingdom claims, be described as "inverse direct effect" of the provisions of that directive in relation to the quarry owners' (emphasis added).

Squintani and Lindeboom 54
the criteria for the assessment development consent requests. The dispute between Mrs Wells and the public authority must be located in this latter situation. Moreover, it is in this latter situation that the Directive is invoked to change the norm that the public authority must apply to assess the request for development consent. Therefore, only this latter norm is affected by direct effect. The first norm-ie the one requiring the quarry owner to obtain development consent to exploit the quarry-remains unaffected by the direct effect of the Directive. The quarry owner must continue to comply with this rule, just as he had to before Mrs Wells' successful case. The annulment of development consent leads to the resetting of the factual circumstances to the situation which existed prior to the request for development consent.
Abstracting from this case, we can state that direct effect does not create direct obligations for individuals when it affects a norm different from the one imposing a positive or negative obligation upon an individual. In such cases, direct effect leads to a change in the factual circumstances which precede the norms in which an obligation is imposed upon an individual. In other words, it leads to mere adverse repercussions.
The normative impact theory is based on the distinction between the norms directly governing the dispute to be resolved on the one hand, and the factual framework to which these norms are applied on the other. 'Factual' is thus meant in a relative sense: the factual framework consists of all circumstances which are taken to be true for the purposes of the dispute. These factual circumstances always include normative components in themselves. 205 In Wells, for example, the factual circumstances include the fact that the quarry owner needs development consent to exploit the quarry (among many other rules). However, relative to the dispute under litigation, these normatively-laden circumstances are considered to be the facts to which another norms is applied. 206 As we will show in the following section, the normative impact theory accurately describes the existing case law. Moreover, recently, in Smith, the Court clarified its case law on the distinction between 'direct obligations' and 'mere adverse repercussions' in a manner that appears to endorse our approach-albeit by using different terminology. 207

B. The case law on direct effect through the lens of the normative impact theory
The normative impact theory reconciles the alleged inconsistencies in the case law of the ECJ from the perspective of the theories reviewed in Section IV, while remaining in keeping with the normative validity of the relationship between primacy on the one hand, and legality and legal certainty on the other. Illustrating the normative impact theory first through the 'incidental effects' cases (Section V.B.i), prior to turning to the horizontal and inverse vertical direct effect cases (Section V.B.ii), helps to explain how our approach functions.

(i) The incidental effects case law
In CIA Security, the factual circumstances of the case consisted of the negative publicity circulated by CIA's competitors, Signalson and Securitel. To establish whether these factual actions were in breach of the relevant norm on unfair practices, another factual circumstance needed to be assessed, concerning CIA's alarm system which had to be marketed according to the relevant norm on alarm systems standards. Direct effect plays a role in this second situation. As a result of the direct effect of the Notification Directive, CIA's alarm system complied with the norms on alarm system standards. This results in the factual conclusion that Signalson and Securitel's claims as regards CIA's conduct were untruthful. Therefore, based on the norms of unfair commercial practices, they can be compelled to cease their practices. As we can see, direct effect did not alter the norm imposing this obligation. It only had an effect on the factual circumstances which preceded the assessment of Signalson and Securitel's behaviour under the norms on unfair commercial practices.
Similar constructions can be found in Unilever v SmithKline and Unilever Italia. The first case concerned an action for an injunction sought by Unilever against SmithKline for misleading advertising, due to statements on toothpaste tubes which, according to Unilever, violated national cosmetic law. SmithKline, by contrast, contended that national law violated Council Directive 76/768/EEC. 208 Invoking this Directive had the effect of disapplying the relevant national law, with the result that the statements on toothpaste tubes were not contrary to any applicable law. This fact had the mere adverse repercussion for Unilever that its action for injunction had no basis in law. 209 Similarly and more well-known, Unilever Italia concerned a dispute which was adjudicated under Italian contact law, but where the answer to the legal question related to the contractual dispute depended on the factual question whether the olive oil of olive oil. It is in the latter circumstance that direct effect plays a role-not on the applicable rules on the law of contract.
Another unruly horse to other doctrinal constructions of the Court's case law, Ruiz Bernáldez, can be explained by the normative impact theory as well. It should be noted that Ruiz Bernáldez was limited to questions on the compatibility of national law with Directive 87/102/EEC. 210 From a scientific point of view, the case is better conceived as involving questions on the compatibility of national law with an EU directive. 211 However, the normative impact theory illustrates that the outcome of the case would have been no different had an explicit question on the invocability of the relevant Directive been asked. In Ruiz Bernáldez, the victim of a car accident caused by an intoxicated Mr Ruiz Bernáldez claimed damages from Ruiz Bernáldez's motor vehicle insurance company. The insurance company relied on national law allowing it to exclude damage caused while intoxicated from the scope of motor vehicle insurance contracts. Such exclusion clauses are prohibited by Article 11 of Directive 87/102/EEC. Accordingly, invoking this Directive modified the norms governing the contractual relationship between Mr Ruiz Bernáldez and his insurance company. From the perspective of the dispute between the victim of the car accident and Ruiz Bernáldez's insurance company, the question of whether the insurance contract covers damage caused under intoxication is of a factual nature. Once it is established that, as a matter of fact, Ruiz Bernáldez had an insurance covering the kind of damage he caused, Mr Ruiz Bernáldez's insurance company was obligated to pay compensation. As such, Ruiz Bernáldez is similar to CIA Security and Unilever Italia. In other words, direct effect modified the norms concerning the contractual relationship between Ruiz Bernáldez and his insurance company, which was subsequently considered to be part of the factual circumstances that are taken to be true in the context of the civil liability case between Ruiz Bernáldez and his insurance company, on the one hand, and the victim, on the other. The fact that the insurance company is obligated to cover damages incurred by Ruiz Bernáldez in this civil liability case is a mere adverse repercussion of invoking the Directive in the preceding contractual relationship.
Ostensibly, the recent Smith case reveals a similar factual situation. As the rear of Mr Meade's van was not designed and constructed with seating accommodation, Meade's motor insurer FBD refused to provide him an indemnity on the basis of its insurance policy, which was in conformity with Section 65(1)(a)(i) of the Irish Road Traffic Act, defining persons not sitting in seating accommodation as 'excluded persons' not having a right to insurance compensation. While Mr Smith relied on Article 1 of Directive 90/232/EEC to claim compensation, the ECJ held that this would entail horizontal direct effect against FBD, thus rejecting Smith's claim. 212 The Court distinguished Smith from Ruiz Bernáldez by pointing out that the question of direct effect was not addressed in the latter case. 213 Indeed, as we will show in Section VI below, the functioning of the preliminary reference procedure significantly affects the relevance of individual cases for the doctrine of direct effect. Nevertheless, in addition to this argument, the normative impact theory suggests there is also a substantive difference between Ruiz Bernáldez and Smith that legitimizes the different outcomes. As observed above, in Ruiz Bernáldez there is a clear demarcation between the norms governing the contractual relationship between Ruiz Bernáldez and his insurance company (the exclusion of coverage for damage caused while being intoxicated), and the subsequent norms governing the civil liability case between Ruiz Bernáldez and its insurance company, on the one hand, and the victim, on the other (the existence of an obligation to compensate the victim). In Smith, no such demarcation exists because the exclusion clause did not merely define the contractual relationship between Meade and FBD, but also directly addressed Mr Smith himself by excluding a specific category of victims from the scope of the obligation to compensation. Using the Wells formula, it is clear that in this case invocability results in horizontal direct effect: invoking the Directive to remove the clause excluding an obligation to compensate the damage of Mr Smith is directly linked to the performance of the obligation to compensate Mr Smith, which is precisely the obligation pursuant to Article 1 of Directive 90/232/EEC.
A peculiar set of facts in relation to direct effect can be seen in Pafitis. In this case, Article 25 of Directive 77/91/EEC was invoked by the old shareholders of a Greek bank against a capital increase which had been independently decided by the temporary administrator of the bank. Article 25 required that any increase in capital was decided upon by the general meeting of shareholders. The Court held that Article 25 of the Directive could indeed be invoked by the old shareholders, as a result of which two types of negative consequences can be identified. The new shareholders lost their shares, which should be considered a mere adverse repercussion of the nullity of the capital increase. Akin to the mere adverse repercussions in Wells, the nullity of the new shareholders' shares was an automatic, factual consequence of the successful invocability of the Directive against the decision of the temporary administrator of the bank. In contrast, the Directive directly imposed an obligation upon the bank and its temporary administrator to annul the capital increase. However, looking closely at the facts of this case, it becomes apparent that the Greek bank concerned cannot be considered an ordinary private entity. The Greek bank had been put under the supervision of a temporary administrator. Under Greek law, this supervision was possible if 'such action is necessary and urgent in order to safeguard the interests of the State, the bank or third parties, or to forestall possible adverse repercussions on the financial market or the economy in general'. 214 Accordingly, as known by the ECJ, 215 it is clear that the Greek bank itself could no longer be regarded as a private party, but should be considered part of the Greek State under the Marshall and Foster case law. 216 The direct obligations for the bank as a result of the invocability of Directive 77/91/EEC thus represent vertical direct effect, while the consequences for the new shareholders were mere adverse repercussions.
(ii) The case law affirming the prohibition of horizontal and inverse vertical direct effect The difference between obligations and mere adverse repercussions becomes clearer when the case law discussed in the previous section are compared to Faccini Dori and Pfeiffer. In Faccini Dori the legal question under scrutiny by the judge was whether Mrs Faccini Dori should pay for the language course she had bought at Milan central station. The answer to this question is not dependent upon the evaluation of any preceding circumstances. The fact that Mrs Faccini Dori bought a genuine language course was not disputed, and it was not disputed either that this transaction was valid and complied with the law on selling techniques, possible product standards, etc. A directive was invoked in this case to change the norms of Italian contract law. It is only by changing the rule that a contractual party has to pay for the performance delivered by the other contractual party that Mrs Faccini Dori could escape contractual liability. Were the directive to apply, the language course seller would have had to refrain from providing the language course (or would have to take it back) and would have to refrain from seeking payment for it. This would have been in every way a case in which direct effect changes a norm to impose an obligation upon an individual. 217 There is a similar construction in Pfeiffer. This dispute concerned a situation where an employer requires its employees to work longer than 48 hours weekly. The resolution of this situation depended on the applicable norm on weekly working hours. No other circumstance needed to be assessed prior to assessing the one in dispute in Pfeiffer. That the employee had a contract for work and got paid for the hours worked was not in dispute. Neither in dispute were the norms concerning the establishment of working relationships and the payment of working hours. Direct effect in this case was invoked to change the norm on the number of hours which may be worked weekly. It was only by changing this rule that Mr Pfeiffer and others could force their employer to reduce the number of hours they were expected to work each week. Like Faccini Dori, in this case direct effect would have changed the norm to impose an obligation upon an individual in a dispute before a national court.
In the inverse direct effect cases the relationship between direct effect and the obligations upon individuals is even more straightforward. In Kolpinghuis, Berlusconi, Arcaro, and Pretore di Saló, the imposition of obligations upon individuals depended on the norm used to qualify their behaviour. In Berlusconi, regardless of whether direct effect was allowed, the applicable norm for resolving the case was in any case a norm from national law. In contrast, in Kolpinghuis direct effect would entail application of the directive itself, ie a substitution effect. Nevertheless, in both cases, direct effect of the respective directives would directly impact the norms to be applied to solve the cases themselves, regardless of whether that norm is of national law, such as in Berlusconi, or of the relevant directive, as in Kolpinghuis. What matters is that, in all these cases, direct effect would have changed the norm to impose obligations upon individuals. This shows how the normative impact theory differs from the substitution-exclusion theory and, as a result, elucidates judgments that remain otherwise obscure.
Somewhere in between Wells, CIA, Unilever, on the one hand, and Faccini Dori, Pfeiffer, and the criminal law proceedings, on the other hand, we have El Corte Inglés. Indeed, although it seemed like a consumer case similar to Faccini Dori, this case presented a somewhat more complex situation. The dispute between El Corte Inglés and Mrs Blázquez Rivero for the reimbursement of the credit loan given by the former to the latter followed a dispute about the non(-adequate) performance of a travel contract between Mrs Blázquez Rivero and Viajes El Corte Inglés SA. This case is thus more similar to CIA Security and Unilever than Faccini Dori and Pfeiffer, with one important difference. In El Corte Inglés direct effect was not invoked to qualify the performance of the travel contract. It is invoked instead to hold the finance company accountable. Clearly, here direct effect would have modified the norms establishing the liability of an individual, leading thus, if allowed, to the direct imposition of an obligation.
In administrative law, similar constructions could also be possible. A good example of such a situation would be a case in which a party requests a public authority to undertake enforcement action against a private party due to a breach of an unimplemented obligation stemming from a directive. Take for example the Environmental Liability Directive. 218 It requires operators of economic activities, business, or undertakings, to take certain measures to prevent or remediate environmental damage. 219 In the absence of national measures transposing the Directive, a party cannot force the competent authority to rely on Article 5 or 6 of the Directive against an operator who failed to comply with any of these provisions. If this were possible, it would imply the imposition on the operator of an obligation stemming directly from the Directive. 220 Indeed, direct effect would in this case determine which norm had to be imposed upon the operator, as also occurred in Kolpinghuis, Berlusconi, Arcaro, and Pretore di Saló. Whether the enforcement measures are explicitly mentioned in the EU provisions at hand, as in the case of Article 4(5) of Directive 1999/22/EC relating to the keeping of wild animals in zoos, 221 is irrelevant in this regard. 222 If invoked, they would change the norm to impose an obligation upon individuals, hence entail the direct imposition of obligations from a directive upon an individual. A conjunctive perspective on both Berlusconi and (hypothetical) cases involving the invocation of provisions from the Environmental Liability Directive shows how the normative impact of invocability must be dissociated from the question of whether the respective provisions includes in its material and personal scope the actions of private parties. Invoking a directive can change the norm directly governing the dispute-entailing the imposition of an obligation onto an individual-regardless of whether the directive is aimed at regulating private party actions. Figure 1 shows graphically the functioning of the normative impact theory in the three scenarios discussed in this section, as well as in Section V.B.i.
In conclusion, this section and Section V.B.i. showed that the normative impact theory is capable of explaining the meaning of the direct obligations versus mere adverse repercussions dichotomy and of reconciling all cases in which the Court explicitly discussed the issue of the invocability of direct effect of directives. It therefore has descriptive accuracy. Moreover, it has normative desirability as it reconciles primacy with legality and legal certainty in a more refined manner than the substitution-exclusion theory, thereby decreasing the risk of false negative outcomes in allowing for the invocability of directives. Conceiving direct effect as invocability, the normative impact theory more accurately grasps the normative effects of invoking a directive on the applicable legal framework(s) and the question of what it means to impose an obligation on to an individual. 219 Arts 5 (on prevention) and 6 (on remedying environmental damage). 220

VI. The preliminary reference procedure and the borderline between direct and indirect effect
A possible critique to our claim that the normative impact theory has descriptive accuracy could be based on two cases in which the Court, while not explicitly discussing direct effect, held that national law was not compatible with the invoked directives, namely Bellone and Stadt Wiener Neustadt. Following the normative impact theory, both cases would have to be qualified as involving, Figure 1 Graphic representation of the normative impact approach* *Ruiz Bernáldez is not mentioned in the figure as we do not consider it a case of direct effect Squintani and Lindeboom 62 respectively, horizontal and inverse vertical direct effect of the relevant directives. However, this critique fails to appreciate the relevance of the preliminary reference procedure on the discussion on horizontal direct effect of directives. As we will show in this section, Bellone and Stadt Wiener Neustadt suggest the existence of a presumption of the possibility of consistent interpretation insofar as the preliminary reference does not address the question of direct effect itself. This interpretation, supported by the Court's judgment in Centrosteel, points at the intricate borderline between indirect and direct effect, and the Court's 'stoneby-stone approach' to developing its jurisprudence. 223 Implicit support is also given by Smith, where the Court recalled the absence of an obligation under EU law to give horizontal direct effect to directives when it is established that 'the national court finds itself unable to interpret provisions of its national law in a manner that is compatible with a directive'. 224 We will illustrate the operation of this presumption in Section VI.A. In Section VI.B, we further discuss a related issue of the functioning of the preliminary reference procedure, namely the manner in which the national court formulates the relevant national legal framework as well as its questions. To this end, we explain how the Court might have reached a wrong conclusion in Ruiz Bernáldez, had the issue of direct effect been raised explicitly by the national court. Lastly, in Section VI.C we conclude that the Court should not be afraid to address the dimension of direct effect explicitly even if it is not explicitly raised by the national court. This would avoid misunderstandings related to the borderline between direct and indirect effect, the presumption of consistent interpretation, as well as the distinction between direct obligations and mere adverse repercussions, without threatening the undeniable truth that the Court tends to solve only one case at a time.

A. The possibility of consistent interpretation
The relevance of the preliminary reference procedure for the interpretation of cases seemingly concerning direct effect was observed by some commentators on incidental effects case law. They observed that in Bellone, as well as Ruiz Bernáldez, the referring national court did not ask whether the litigant could invoke a directive against another individual, arguably restricting the scope of the reference to the compatibility of national legislation with the applicable directives. 225 While it is true that Ruiz Bernáldez is likely to be qualified as a 223 On the 'stone-by-stone' approach, see Lenaerts (n 11 Bellone concerned a contractual dispute between Mrs Bellone and Yokohama SpA. Mrs Bellone acted as commercial agent on behalf of Yokohama pursuant to an agency contract. 226 After Yokohama terminated this contract, Bellone claimed payment of various indemnities. These claims were initially rejected on the ground that the agency contract was void, as Mrs Bellone was not registered in the register of commercial agents, which was a compulsory contractual form on the basis of Article 9 of Italian Law No 204. On appeal, Mrs Bellone invoked Article 1 of Directive 86/653/EEC which defines 'commercial agent' by reference to the activity pursued and does not require any other registration measures. 227 The Court indeed confirmed that the registration requirement in Italian law was incompatible with the Directive. 228 The direct effect of Article 1 through either an exclusionary effect or a substitution effect 229 would directly impact the norms governing the dispute between Mrs Bellone and Yokohama, since the disapplication of Article 9 directly confirms the existence of an agency contract between Mrs Bellone and Yokohama, thus determining the question of whether Mrs Bellone had any right to indemnities. Under the normative impact theory, the case involves a direct obligation. Nevertheless, the Court ignored the horizontal nature of the dispute.
Subsequently in Centrosteel, shortly after Bellone, another Italian national court referred an identical question to the Court with reference to Bellone. The referring court noted that Bellone could not result in Law No 204 being disapplied in the proceedings before it, since this would entail a horizontal direct effect of Directive 86/653/EEC. The Court in turn observed that indeed directives cannot of themselves impose obligations on individuals. But, the Court emphasized, the case law also requires national courts to interpret their national law as far as possible in light of the wording and purposes of applicable directives. It continued: As the Advocate General points out in paragraph 36 of his Opinion, it seems in that regard that the Corte Suprema di Cassazione (Supreme Court of Cassation), following the judgment in Bellone, has changed its case-law so that a failure to comply with the obligation prescribed by Law No 204 to be entered in the register of commercial agents and representatives no longer entails the nullity of an agency contract in Italian répondre aux questions posées, sans attirer l'attention des jurisdictions nationales sur l'exclusion de l'effet direct horizontal'. See also, van Gerven (n 51), 259f and Stuyck (n 51). 226 Bellone (n 51), paras 2-7. 227  Hence, the Court emphasizes the link between the incompatibility of national law with the Directive and the obligation of national courts to remedy this incompatibility by interpreting the national provisions applicable, so far as possible, in light of the Directive. 231 In the absence of any indication provided by the national court that this incompatibility cannot be remedied through consistent interpretation, the Court indeed appears to continue under the presumption that consistent interpretation is possible. Accordingly, Centrosteel shows that Bellone is not relevant to the development of the doctrine of direct effect and thus is not an anomaly in the case law.
A similar conclusion can be reached as regards Stadt Wiener Neustadt in the context of a triangular situation and a possible inverse vertical effect. Like Wells and Salzburger Flughafen, this case also concerned the invocability of the EIA Directive. However, rather than an action for annulment of development consent granted in breach of the EIA Directive, as was the case in Salzburger Flughafen, Stadt Wiener Neustadt concerned a request for remedial action against the negative effects of a development consent granted in breach of the EIA Directive requirements, which had become definitive under national law. In 2014, a request by the Environmental Ombudsman to determine whether the plant should be subjected to an EIA in accordance with the UVP-G 2000 was rejected on the basis of Paragraph 46(20)(4) UVP-G 2000, stating that projects the consent for which is no longer at risk of annulment, as in this case, must be regarded as having been authorized in compliance with the requirements for an environmental impact assessment. The ECJ considered whether Paragraph 46(20)(4) UVP-G 2000 could be invoked to reject a request for remedial action against the failure to perform an environmental impact assessment, as made by the Environmental Ombudsman, and found in the affirmative.
Unlike Wells and Salzburger Flughafen, in this case the negative consequences of this obligation for national authorities for the operator of a waste treatment facility will not flow automatically from the decision of the national court to annul the negative reply of the Government of the Land to the request by the Environmental Ombudsman. Once this decision is annulled, the Government of the Land will need to decide whether the project had to be subjected to an EIA. The outcome of the assessment by the Environmental Ombudsman will be a new, autonomous decision, and this decision could be that a full-fledged EIA must be performed. If an EIA must be performed indeed, and paragraph 46 of the Court's ruling hints in this direction, the consent could be suspended or withdrawn while waiting the results of the impact assessment. 232 The question of how the content of this new, autonomous decision can be squared with the prohibition of inverse vertical direct effect was not part of the preliminary reference, and the Court paid no attention to this matter at all. However, if we consider Stadt Wiener Neustadt in light of the Court's explanation of Bellone in Centrosteel, it becomes clear that the question of the direct effect of the EIA Directive is not only not explicitly discussed in the judgment, but also that the question of direct effect is not (yet) relevant as long as there is a presumption that consistent interpretation is capable of remedying the conflict at hand. Hence, in the absence of a specific confirmation by the referring national court that consistent interpretation is impossible, the Court proceeds on the presumption that the incompatibility of national law with the EIA Directive can be solved through consistent interpretation. 233 Under this presumption the question of direct effect is to be considered a purely hypothetical question which lies outside the scope of the preliminary reference procedure. 234 B. The formulation of the relevant national legal framework and the preliminary question The second manner in which the functioning of the preliminary reference procedure has affected the case law on direct effect of directives is the manner in which the referring court presents the relevant national legal framework. Again, Ruiz Bernáldez in conjunction with Smith provides an illustrative example. As noted above, there are two potential explanations as regards the analytical relationship between these two cases. In Smith, the Court distinguishes Smith from Ruiz Bernáldez by observing that the scope of the preliminary reference in the latter case was confined to the compatibility of national law with EU law. 235 It suggests, therefore, that had the referring national court in Ruiz Bernáldez specifically raised the question of direct effect, the Court would have answered that the relevant Directive may not be invoked because of the prohibition of horizontal direct effect. By contrast, using the normative impact theory, we hypothesized in Section V.B.i. above that the phrasing of the relevant national legal framework in Ruiz Bernáldez was such that invoking the Directive would not lead to direct obligations for the insurance company, because the Directive only governs the contractual relationship between Ruiz Bernáldez and his insurance company, not their relationship with the victim of the car accident.
However, looking at the manner in which the national court had referred the case to the ECJ, we admit that it is unlikely that the Court would have allowed for direct effect in response to a hypothetical question to this end. In its reference, the national court presented to the Court Article 12(3)(b) of the Reglamento del Seguro Obligatorio, which, as noted above, excludes from insurance coverage all damage that is caused by an intoxicated driver. The national court, in doubt as to whether this article is in conformity with Community directives relating to insurance against civil liability in respect of the use of motor vehicles, was not certain whether: Article 12(3)(b) of the Reglamento del Seguro Obligatorio could be interpreted as meaning that the insurer did not have to compensate the victim of a road-traffic accident caused by an intoxicated driver. 236 The emphasis added to this section is to illustrate that in the preliminary reference, the national court focuses on the actual result of applying the Directive. This result of applying this Directive is that the insurance company would not have to indemnify Mr Ruiz Bernáldez for damages liable to the victim. In practice, of course, this would mean that the insurance company would not have to compensate the victim. Assuming that this result would not have been possible through consistent interpretation, 237 the question still remains whether the insurer having to compensate the victim is a direct obligation or a mere adverse repercussion. However, emphasizing again the ultimate consequence of direct effect, the actual question asked by the national court was: If a statutory provision or contractual clause which excludes insurance cover where the driver responsible for the damage is intoxicated is valid in relations between the insurer and the insured, could its validity as against a third party who has suffered harm be considered to be in compliance with the system laid down in Directives 72/166/EEC, 84/5/EEC and 90/232/EEC? 238 Also here, the national court emphasizes the relationship between the insurer and the victim. A direct effect of the Directive would clearly affect this relationship, suggesting a horizontal direct effect. It is not inconceivable that the Court, had it based a hypothetical answer to a question of direct effect on this phrasing of the preliminary question, would have concluded that invoking the Directive was not allowed in this case, for it would have entailed an obligationthat otherwise did not exist-on the insurer to compensate the victim. Only by very precisely looking at the relevant national legal framework and the norms directly affected by the invocability of the Directive can one distinguish the direct effect of invoking the Directive (the altered insurance relationship between Ruiz Bernáldez and his insurer), and the miscellaneous consequences that this entails (the indemnification of Ruiz Bernáldez for damages liable to the victim).
This conclusion may suggest that the normative impact theory is also too formalistic-just like the substitution-exclusion theory-in that the question of whether a directive can have direct effect depends at least in some cases on the way in which the national legal framework is formulated, and how the national court presents it to the ECJ. However, unlike the substitution-exclusion theory, the normative impact theory remains better in keeping with the distinction between imposing direct obligations and entailing mere adverse repercussions. The substitution-exclusion theory necessarily produces false positive and false negative outcomes in light of the rationale of the prohibitions of inverse vertical and horizontal direct effect: as shown above, for example, invocabilité d'exclusion is clearly capable of entailing obligations that are a direct normative consequence of invoking a directive. 239 By contrast, the normative impact theory is precisely premised on the distinction between direct obligations and mere adverse repercussions that are indirect.
Of course, this theory is as formulaic as any other; 240 the reach of direct effect has always been 'l'art du possible', as Pescatore has already observed. 241 Nevertheless, we submit that the normative impact theory accurately describes what is possible in terms of the horizontal direct effect of directives. Inevitably, 238 Ruiz Bernáldez (n 51), para. 5 (emphasis added). 239 See Section IV.C. above and Berlusconi (n 42) in particular. 240 See also n 192 and accompanying text. 241 Pescatore, 'The Doctrine of "Direct Effect"' (n 1), 177. this creates the situation where the formulation of the relevant national legal framework (both in the law itself and how the national court presents it to the ECJ), at least in some cases, becomes essential to the question of whether a directive can be invoked in a horizontal or an inverse vertical dispute. 242 This makes it even more important that there is clear communication between national courts and the ECJ in terms of the scope of the preliminary reference procedure and the manner in which national law is formulated, but also the scope of the answer of the ECJ. In light of the limited information that is available to the Court, its judgments in preliminary reference cases must be scrutinized carefully to deduce not only what the Court says, but in response to what it is saying this, and thus how its judgment should be understood.
C. Deciding one case at a time (but not less) As Lenaerts and Corthaut observe, the preliminary reference proceedings increasingly reveal the impact of invocability questions on the outcome of individual cases. 243 As the Court considers itself bound by the scope of the specific questions asked by the referring court, 244 it is not unreasonable for it not to fully take into account all possible follow-up situations which may or may not arise after its judgment is applied by the referring national court. In this regard, it is logical to presume that consistent interpretation is a viable option for the national court as the first and normatively preferred mechanism to remedy deficits in the proper implementation of EU directives. Moreover, the Court is clearly applying a 'one case at a time' method of adjudication 245 -or in other words, a 'stone-by-stone approach' 246 -which explains the absence of dicta that are not directly relevant to the preliminary questions asked by the referring national court.
On the other hand, precisely because the preliminary reference procedure is the fundamental communication mechanism between the national and the EU courts, presuming that the logical consequences of a preliminary judgmenteven if formally outside the scope of the Article 267 procedure-must be dealt with by the national court without guidance by the Court is not without risks. 247 The same applies to the presumption that any follow-up issues, 242 Another case where the formulation of the national legal framework is essential to the question of whether direct effect is possible, also in conjunction with the possibility of consistent interpretation to solve the incompatibility concerned, is Stadt Wiener Neustadt (n 7). 243  including not least the possibility of interpreting national law in conformity with EU law, are not to be addressed by the Court right away, but are to be addressed by the national court, potentially leading to another preliminary reference procedure. 248 Moreover, sometimes it is not unreasonable to conclude that the Court does not even solve a full case at a time, which is damaging to the preliminary procedure, trust among courts, and the effectiveness and uniform application of EU law. 249 For these reasons, it is advisable that the Court should be more careful in handling cases (potentially) having an invocability dimension. It certainly would not hurt to emphasize obiter dictum the obligation of national courts to remedy incompatible national law through consistent interpretation, as well as the wellestablished case law on the limits to direct effect. 250 This would also avoid misunderstandings of the Court's case law to the extent that the information supplied by the national court might incline the Court to reach conclusions that are inconsistent with its own case law. As Ruiz Bernáldez illustrates, the manner in which national courts phrase their references might-accidentally-entail a wrong conclusion on part of the Court. Moreover, because the presumption of the possibility of consistent interpretation has never been explicitly acknowledged by the Court as such, right conclusions on part of the Court may nevertheless be wrongly interpreted by legal academics, national courts, and practitioners, leading to an unnecessary avalanche of legal commentary and widespread confusion in the legal community. 251 Being more explicit on all the legal principles and doctrines that apply to the national dispute at hand would help to avoid such (potentially far-reaching) misunderstandings.

VII. Conclusions
Direct effect cannot be seen as an infant disease of EU law, or merely a matter of justiciability. As long as the EU legal order and the national legal orders are not fully assimilated, and EU norms are to a large extent enforced through the administrative and adjudicative fora of the Member States, further theory-building on the invocability of EU law remains necessary.
This article has provided a first step in this direction through four claims. First, after providing a bird's eye view of the development of direct effect case law and the main theories that have tried to theorize direct effect in general and of directives specifically, we clarified the case law on the direct effect of directives in inverse vertical, triangular, and horizontal disputes. We did so by showing that the distinction between 'direct obligations' and 'mere adverse repercussions' holds the key to a proper understanding of this jurisprudence. In other words, this distinction ought to be seen as the central demarcation line between admissible and prohibited instances of invocability of directives.
Secondly, we showed how the various theories on the direct effect of directives-as summarized in Section II-fail to appreciate the distinction between 'direct obligations' and 'mere adverse repercussions'. Consequently, although some theories have more explanatory power vis-à-vis the case law than others, none of them is capable of providing a watertight description of all the relevant judgments in horizontal and inverse vertical situations. Even in their most accurate formulations, all theories are ultimately based on conceptions of the distinction between 'directly imposing an obligation' and 'entailing mere adverse repercussions' that are irreconcilable with the case law and its rationale.
Thirdly we offered a descriptively accurate and, arguably, normatively desirable doctrinal construction of the case law. According to our normative impact theory, the limits to the direct effect of directives are defined by the impact of invoking a norm of a directive on the norms directly governing the dispute. Thus, the prohibitions of inverse vertical and horizontal direct effect of directives are conceptualized as a prohibition to invoke a norm from an unimplemented directive insofar as this would change the norms which directly apply to the dispute at hand. This is because such a change in the applicable legal framework entails a direct obligation onto the (other) individual. Whether this change is the result of substitution or exclusionary effects is, in this respect, irrelevant. In contrast, invoking a norm from an unimplemented directive does not create a direct obligation for another individual insofar as this invocation affects a norm different from the one imposing a positive or negative obligation upon this individual. In such cases, direct effect leads to a change in the factual circumstances which precede the norms in which an obligation is imposed upon an individual. In other words, it leads to mere adverse repercussions.
This approach reconciles the alleged inconsistencies in the case law of the ECJ from the perspective of the substitution-exclusion theory, while remaining in The Normative Impact of Invoking Directives keep with the normative validity of the relationship between primacy, on the one hand, and legality and legal certainty, on the other. Not only does the normative impact theory succeed in explaining the relevant case law of the Court-including CIA Security, Unilever, Pfeiffer, and Berlusconi-we submit that it is also normatively more sound than existing theories such as the substitution-exclusion theory.
Lastly, we situated this doctrinal construction in the context of the preliminary reference procedure to explain the allegedly remaining anomalies in the case law. The question of direct effect ought to be explicitly raised by the referring national court so as to be part of the preliminary reference procedure. Moreover, the Court appears to proceed on the basis of a presumption of the possibility of consistent interpretation in the absence of concrete evidence to the contrary. Whilst this is understandable from the perspective of the functioning of Article 267 TFEU, this presumption has not been formulated explicitly by the Court to the detriment of doctrinal clarity. Nonetheless, the presumption of the possibility of consistent interpretation is essential to elucidate the existing case law and, more importantly, to understand the scope and meaning of future judgments in cases having an (implicit) direct effect dimension.
The importance of a comprehensive theory on the invocability of EU law transcends the case of directives. We urge the Court to discuss the limits to the invocability of EU norms explicitly. The normative impact theory may help to address these issues in a discernible manner while keeping in touch with the Court's stone-by-stone approach.