Centros and the Internal Market

Abstract

How special is Centros? This contribution places Centros in internal market law. It starts by turning the judgment on its head and imagines an alternative Centros: the judgment that the Danish authorities had wished for and which the Court declined to hand down. The aim of this exercise in legal fiction is to show that the Court could have decided Centros in another way. The choices underpinning Centros are then contrasted with other internal market cases to conclude that both the real Centros and the alternative Centros are in some ways aligned with the free movement case law. It is argued that Saydé’s analysis of abuse best helps to make sense of this confusing state of play. The article ends with suggestions for further research: taking the cost dimension into account in a behavioural perspective could shed light on the allocation of regulatory powers between home state and host state.

1 Introduction

How special is Centros?¹ I will not consider this question from the point of view of the Court’s case law on the mobility of corporations, which has been extensively analysed.² In that regard, it is well established that Centros opened the door to both greater corporate mobility in the internal market and more regulatory competition among Member States. With the limited exception of restrictions on the right of companies to ‘exit’ their Member State of incorporation, which existed before Centros and still remains after Centros,³ the Court remained faithful to the overall very liberal stance to the regulatory mobility of companies it had initiated in Centros.⁴ Corporate mobility motivated solely by the desire to benefit from a more favourable regulatory regime is protected, in the words of the Centros court, as ‘inherent in the exercise of the freedom of establishment’.⁵Centros’ progeny has been extensively discussed from the point of view of company law and private international law. In this contribution, I would like to address the issue of how special a ruling Centros is from the point of view of internal market law.⁶

Centros raises a question which is relevant to all four freedoms of movement and to citizenship: to what extent does EU law allow Member States to fight ‘abuses’ (in the lay sense of the term)? In other words, how much regulatory space do Member States enjoy to discourage or neutralise regulatory mobility (mobility for the sole aim of securing a regulatory benefit)?⁷

To be sure, exercising EU rights strategically is not unique to the freedom of establishment or to companies. Students may opt to study veterinary science in a Member State where admission conditions are less stringent than in their own (Bressol and Chavelot).⁸ A Chinese mother may elect to give birth in Northern Ireland where her baby will be entitled to European citizenship, giving rise to a derived right of residence for the parents (Chen).⁹ A German national named Nabiel Bagdadi at birth may have his name changed (in several instalments) to Peter Mark Emanuel Graf von Wolffersdorff Freiherr von Bogendorff while he resides in the UK and then seek to have his new name recognised in Germany, thus circumventing the abolition of tokens of nobility in the German Constitution.¹⁰ This last example illustrates that cases concerning the free movement of (natural) persons can also concern ‘U-shaped’ situations. Like in Centros, the mobile person ‘moves’ from a home state with stringent rules (on minimum capital or tokens of nobility) to a more friendly host state (the UK in both cases) and then comes back home and seeks recognition of the situation lawfully constituted in the host state (which has artificially been made a home state for the purposes of the second leg of the journey).

Private strategizing can seek to circumvent national rules that would normally apply (as in Centros or Wolffersdorff von Bogendorff). It can also aim to benefit from rules that would not normally apply (as in ‘benefit tourism’ cases). Both can be described as forms of abuses: circumventing the law that would normally apply is the essence of the abuse of right,¹¹ while artificially securing a regulatory benefit characterises an abuse of law.¹² In this regard, Saydé helpfully characterises an abuse of right as an improper exercise of a right and an abuse of law as improper acquisition of a right.¹³ Many authors have considered both forms of ‘abuse’ together, especially when reflecting on the existence of a general prohibition of abuse in EU law.¹⁴ Whetherill notes in this regard that ‘the essence of an allegation of abuse [of either type] is to say that host state control should be permitted to apply without restraint’.¹⁵ Such a broad understanding is helpful for the present purpose of placing Centros in internal market law, and singularly for a comparison with the case law on the free movement of natural persons.

What is puzzling is how differently the Court seems to address the issue of the abuse of rights across cases.¹⁶ In this regard, Centros prima facie appears to be a special case in that the Court seems to leave no room at all for Denmark to neutralise strategic mobility,¹⁷ while, in other cases, it accepts that a (true) home state can neutralise evasion strategies. Examples of this friendlier stance towards states whose laws are being circumvented can be found not only in cases involving tax law (Cadbury Schweppes, Halifax)¹⁸ but also in cases concerning audio-visual regulation (TV10)¹⁹ and the free movement of persons.²⁰

Given these oscillations in the case law, it is noteworthy that, in Centros, the Court presents the solution as the inevitable consequence of a literal reading of Articles 49 and 54 TFEU: By incorporating Centros in the UK, Mr and Mrs Bryne, two Danish nationals, exercised the right to create a company in another Member State, a right conferred by the Treaty upon all nationals of a Member State. In turn, Centros, a Limited Company lawfully incorporated in the UK, has exercised its right, equally guaranteed by the Treaty, to set up a branch in Denmark.²¹ As with all well-drafted judgments, the reader is, at least momentarily, under the impression that such is indeed the implacable logic of the Treaty rules. However, it is also the case that, when the Centros judgment was handed down, it was greeted not only with a mix strong criticism and applause but also with genuine surprise.²² This strongly suggests that, for many, the Court’s reading of the Treaty was not self-evident.

This is why, in Sect. 2, I start by turning Centros on its head and imagine an alternative Centros: the judgment which the Danish authorities had hoped for and which the Court declined to hand down. The purpose of this exercise in juridical fiction is to show that EU law offered all the resources that the Court would have needed to hand down a very different judgment. Next, in Sect. 3, I ask whether Centros or the alternative Centros display a better fit with internal market law. Having thus highlighted inherent ambiguities in internal market law regarding the regulatory leeway that Union law leaves Member States to combat perceived abuses, Sect. 4 turns to doctrinal suggestions of how to make sense of this confusing case law. Section 5 briefly concludes that Centros is extreme rather than utterly special and suggests further potential explanations for the ebb and flow of the case-law governing the lawfulness of strategic mobility.

2 An Alternative Centros

Let us start by turning Centros on its head and imagine the judgment which the Danish authorities had hoped for and which the Court declined to hand down. The reasoning would have unfolded in three steps. First, the Court would have recalled its established case law on the function of freedom of establishment, from which it derives a teleological interpretation of the notion of establishment.²³ To this effect, the Court might have cited Gebhard (handed down 4 years prior to Centros), where it held that

[t]he concept of establishment within the meaning of the Treaty is […] a very broad one, allowing a Community national to participate, on a stable and continuous basis, in the economic life of a Member State other than his State of origin and to profit therefrom, so contributing to economic and social interpenetration within the Community in the sphere of activities as self-employed persons.²⁴

It could also have referred to Reyners, where it had first made clear that establishment means real economic integration in the host Member State. In this judgment, the Court had referred to the Treaty provisions on establishment (together with implementing directives) as

a set of provisions intended to facilitate the effective exercise of this freedom for the purpose of assisting economic and social interpenetration within the Community in the sphere of activities as self-employed persons.²⁵

Indeed, in the first decades of European integration, it was consensual that establishment covered instances of mobility when a genuine link was created with the economy of the host Member State.²⁶ As explained by AG Darmon in this Opinion in Daily Mail,

[Establishment] means integration into a national economy. Thus, it is not contested that establishment within the meaning of the Treaty involves two factors: physical location and the exercise of an economic activity, both, if not on a permanent basis, at least on a durable one.²⁷

Second, the Court would have discussed how this notion of establishment applies to the facts at hand in Centros. Since it was common ground that the Limited Company incorporated in the UK was a letterbox company with no economic activity in that territory, the Court would have indicated to the referring Danish Court that Centros was not truly ‘established’ in the UK within the meaning of what is now Article 49 TFEU.

Third, it would have drawn the conclusion that, not being properly ‘established’ in the UK, Centros could not avail itself of Article 49 TFEU and did not enjoy a right of secondary establishment in Denmark.

Note that this reasoning does not need to mobilise an explicit doctrine of abuse. It is based on an interpretation of the notion of establishment.²⁸ As several authors have remarked, in relation with such open-textured provisions as the free movement rules in the Treaty, there is no real need for a formal doctrine of abuse.²⁹ An interpretation of the scope of free movement rules suffices to combat private strategic mobility to the extent deemed desirable. Indeed, AG La Pergola, in his opinion in (the real) Centros, recognised this when he wrote: ‘to determine whether or not a right is actually being exercised in an abusive manner is simply to define the material scope of the right in question’.³⁰

Of course, a variation is imaginable and an alternative alternative Centros could have relied on a formal doctrine of abuse. The (imaginary) Advocate General in the case should not have recommended it, for reasons of consistency with the case law in other freedoms of movement (exposed below), but in imaginary opinions also, all options must be considered. Since the chronology of the case law does not represent a constraint in legal fiction, applying the Emsland Stärke or Halifax test could have been considered.³¹ Either would have permitted the conclusion to be reached that the transaction was abusive but the Halifax formula should have been preferred. In that case, the Court, following the Opinion of AG Poiares Maduro, addressed the criticism that a doctrine of abuse creates legal uncertainty.³² To this effect, it replaced the original two-pronged Emsland-Stärke test, which referred to an objective and a subjective element,³³ with a wording which clarified that abuse is an objective notion. The Halifax test instructs courts to look at (1) whether the transaction generates a regulatory benefit (the tax benefit in Halifax) which is contrary to the aim of the provisions which formally apply (the VAT directive in Halifax, Article 49 TFEU in alternative Centros) and (2) whether the main purpose of the transaction (rather than the subjective intent) is to obtain this regulatory benefit (the tax benefit in Halifax).³⁴ If the Court had chosen to apply this test beyond the realm of tax law, it would have named the letterbox company a sham or a ‘purely artificial arrangement’ designed to generate a primary establishment contrary to the purpose of Article 49 TFEU which is to facilitate genuine socio-economic integration. On the facts of Centros, it would not have been difficult to conclude that obtaining this legal result was indeed the primary aim of the UK incorporation. For good measure, the Court (still always unconstrained by chronology) could have referred to Article 54 of the Charter of Fundamental Rights (the prohibition of abuse of rights). Note that, with or without explicit recourse to a doctrine of abuse, the result would have been the same and the typical sanction of abuse would have ensued, namely the choice of law would have been denied and Danish law would have applied to the entity operating in Denmark.³⁵

This exercise in juridical fiction only aims to show that, had the Court wished to rule differently in Centros, it could have done so in two different ways, with or without having recourse to a formal doctrine of abuse. Either way, internal market law contains all of the argumentative resources and interpretive techniques needed.

3 Centros, The Alternative Centros and Internal Market Law

In this section, I consider whether the real Centros or the alternative Centros ruling outlined in the previous section is more in line with internal market law. The alternative Centros, while representing the exact opposite of the real one, is not fanciful and would have been congruent with several lines of existing cases. This section illustrates this claim by looking at free movement cases concerning companies (Sect. 3.1) and natural persons (Sect. 3.2).

3.1 The Alternative Centros and the Free Movement of Companies (Freedom of Establishment and Freedom to Provide Services)

Concerning, first, the jurisprudence on the freedom of establishment, it is already clear from the above that an alternative Centros would have been in line with the tax evasion case law. Cadbury Schweppes, Halifax and OyAA (all Grand Chamber judgments posterior to Centros) are cases in point.³⁶ Like Centros, all three involved instances of regulatory mobility: the parties had designed their operations with a cross-border element for the sole purpose of obtaining a regulatory benefit, namely availing themselves of advantageous tax provisions (rather than company law provisions in Centros). In all three cases, the Court ruled that the Member State whose tax provisions were being circumvented could consider these practices as abusive. It did so on grounds of the teleological interpretation of establishment outlined above (the first variant of the alternative Centros). In Cadbury Schweppes, for example, the Court held that

Having regard to that objective of integration in the host Member State, the concept of establishment within the meaning of the Treaty provisions on freedom of establishment involves the actual pursuit of an economic activity through a fixed establishment in that State for an indefinite period […]. Consequently, it presupposes actual establishment of the company concerned in the host Member State and the pursuit of genuine economic activity there.³⁷

It then went on to rule that the fight against abuse was an overriding reason in the general interest which a Member State could invoke to apply its tax law to a corporate entity created for the sole purpose of evading it.³⁸ Similarly, in Halifax (in the context of the VAT directive), the Court explicitly referred to a ‘principle of prohibiting abusive practices’³⁹ and held in very general terms that

The application of Community legislation cannot be extended to cover abusive practices by economic operators, that is to say transactions carried out not in the context of normal commercial operations, but solely for the purpose of wrongfully obtaining advantages provided for by Community law.⁴⁰

In these and in other internal market cases, alleged abuses do not call for a specific analytical framework.⁴¹ They are dealt with exactly as any other free movement case: anti-abuse measures adopted by Member States create obstacles to the freedom of establishment and, as such, are reviewed by the Court. Each case hinges upon whether the measures at issue may be justified on grounds of overriding reasons in the public interest—the umbrella category in which the fight against abuse takes place—and proportionality. What the tax cases illustrate is that the Court does sometimes recognise that it is legitimate and proportionate for Member States to combat strategic private mobility.

TV10 illustrates the same point in relation to the free movement of services. A television channel which was substantially Dutch, in the sense that it operated in the Dutch language, hired mostly Dutch staff and targeted the Dutch public, had incorporated in Luxembourg to evade the Dutch media law. The Luxembourg-based company relied on the free movement of services to claim the benefit of home state regulation. To neutralise this strategy, the Dutch media regulator had decided to treat this media outlet as though it had been established in the Netherlands (different rules applied to Dutch and out-of-state TV channels). Note that this anti-circumvention strategy is very similar to the one the Danish company registrar had sought to deploy in Centros: looking at the reality underneath the convenient out-of-state establishment, stripping the Luxembourg corporation of its artificial legal cloak,⁴² and giving the situation one that better fitted economic and social reality. Arguably, the measure adopted by the Dutch regulator is more proportionate in that it does not deny market access but the logic is the same: the measure neutralises the convenient choice of law and imposes host state rule. Note further that, as Weatherill underscores, the Court did not have recourse to a formal doctrine of abuse. Rather, it reasoned using the usual obstacle/justification paradigm characteristic of free movement cases.⁴³VT10 is a relatively rare case where the justification of the host state (the true home state, artificially turned host state by the very strategy designed to circumvent the application of its law) is accepted as not only legitimate but also proportionate. Either variant of the alternative Centros would have been in line with TV10 in recognising that a Member State may call the corporate bluff and win.

3.2 The Alternative Centros and the Free Movement of Natural Persons

Some free movement of persons cases, such as Chen or Wolffersdorff von Bogendorff mentioned in the introduction, show that natural persons also make use of free movement strategically. However, it is extremely rare that the free movement of persons involves a purely artificial construct. Because natural persons are not legal fictions, they usually genuinely spend time in the host state of their choosing. They study, work or form relationships there. While it is of course possible that some strategies may appear ‘abusive’ in the lay sense of the term, it is very difficult to find cases that are truly comparable with Centros as far as the artificiality of the situation is concerned. Setting this importance difference aside, it is possible to draw comparisons in two other respects: the treatment of U-shaped situations (which are not artificial) and the requirement of a genuine link.

3.2.1 Non-artificial U-Shaped Situations and the Importance of Harmonisation

The alternative Centros takes the view that, in an artificial U-shaped situation, the strategic circumvention of host state law deserves to be neutralised, just like in TV10. If a neutralisation of the choice of law is the sanction of abuse, then surely such neutralisation should occur only when mobility is exercised artificially. In the absence of artificiality, free movement rights should prevail. To test whether this simple logic guides the case law, it is interesting to consider a non-artificial U-shaped situation involving natural persons. Such a configuration occurs, for example, when a national of a Member State studies abroad and then returns to her home Member State. The degree or work experience acquired during her stay in the host Member State leads to the creation of a legally valid situation in that state, such as the right to exercise a regulated profession. The case law in this area shows that the protection of a state’s regulatory interests goes well beyond allowing national authorities to legitimately combat abuses and is intrinsically linked to the state of harmonisation. The importance of harmonisation can be illustrated by contrasting Knoors with Bouchoucha.

In Knoors,⁴⁴ for example, a Dutch plumber had acquired several years of professional experience in Belgium. His training and experience were considered sufficient to qualify as a plumber in that Member State, but not in the Netherlands, his native country, where a particular certificate was required. Nothing in the case suggests that his stay in Belgium had the sole purpose of circumventing the Dutch professional regulation. Yet the Dutch government raised the issue and the Court reassured it that

It is not possible to disregard the legitimate interest which a Member State may have in preventing certain of its nationals, by means of facilities created under the Treaty, from attempting wrongly to evade the application of their national legislation as regards training for a trade.⁴⁵

On the facts, however, it ruled that a directive had harmonised away the risk of abuse.⁴⁶ In a similar case but in the absence of harmonisation, Mr Bouchoucha, a French citizen, had earned his diploma in osteopathy in London and then returned to France to practice.⁴⁷ Mr Bouchoucha fell foul of French law, which reserved the practice of osteopathy to medical doctors. Again, it can hardly be advanced that staying for several years in a Member State and studying there amounts to an abuse or an artificial arrangement. Yet, there was undeniably a strategic dimension: Mr Bouchoucha went to study osteopathy in the UK because no such training was offered in France. The benefit that he obtained from this was both real (training and a degree) and regulatory (a title recognised in the UK). It is noteworthy that the Court seemed to give more weight to the second aspect than to the first. While it was not disputed that the training was genuine, the French point of view that mobility creates an undue claim to practice osteopathy in France without being a qualified doctor seemed to win the Court’s sympathy. In its judgment, the Court did not invoke the aim of free movement law and the creation of genuine economic, scientific and social ties or the fact that such ties are created by such genuine mobility as that of Mr Bouchoucha. Instead, it reiterated the decision in Knoors cited above and this time ruled that France could apply its professional regulation and request Mr Bouchoucha to attend a medical school if he wanted to practice in France (on pain of criminal sanctions).

Two elements may be highlighted here. First, the fight against abuse as a justification for obstacles to trade seems to have an eminently variable scope. In Centros, this justification received the narrowest imaginable scope. By contrast, its scope is quite broad in Bouchoucha. Second, read in conjunction, Knoors and Bouchoucha suggest that harmonisation is a crucial factor determining the extent to which governments can hope to invoke an anti-abuse justification successfully. Harmonisation works as EU protection against abuses and deprives Member States of the possibility to design their own unilateral anti-abuse measures. Indeed, this is an important background element to understand the surprise that Centros caused. As Gelter explains, in the first decade of the internal market, it was consensual that company law would be harmonised, precisely because the absence of harmonisation would lead to ‘abuses’.⁴⁸ The tacit understanding was that, so long as harmonisation had not progressed, Member States retained a wide margin of regulatory discretion to fight abuses, for example by declining to recognise out-of-state corporations or refusing to grant them equal treatment.⁴⁹ It is this understanding that Centros tore apart: although the harmonisation of corporate law had stalled, Member States now had no room at all for neutralising what many considered to be a wrongful circumvention of their laws.⁵⁰

3.2.2 Centre of Gravity and Benefit Tourism

Besides the treatment of U-shaped situations, the treatment of the ‘genuine link’ requirement offers a further opportunity for a comparison between Centros and the case law on the free movement of persons. The alternative Centros turns on a requirement that a corporation should have a genuine link with the economy of the State where it is incorporated in order to enjoy the rights guaranteed by Article 49 TFEU, a requirement that was squarely rejected in the real Centros.

The law of the free movement of persons offers multiple examples where the enjoyment of an EU right—usually the right to non-discrimination in relation to some benefits—is conditional upon the existence of a genuine link with the host country. Collins provides a first illustration.⁵¹ Mr Collins was a dual US and Irish national. During his student years, he had spent some time in the UK, where he had worked in various small jobs. After a long absence of almost 20 years, he returned to the UK and claimed unemployment benefit as a mobile worker. The Court agreed with the UK government that the passage of time had dissipated any genuine links between Mr Collins and the UK labour market. On this basis, the benefit of equal treatment could lawfully be denied. More generally, Ziegler, studying abuse in the case law on the free movement of workers, concludes that the Court relies on a ‘time and scale element’ to appraise a genuine link.⁵²

The case law on ‘benefit tourism’ provides the clearest and most frequent expression of the compatibility of genuine link requirements with the law of free movement. Such a requirement was developed in the case law on student benefits and is routinely used in cases involving any sort of social benefit.⁵³ For example in Bidar, the Court accepted that the UK may require migrant students to show a genuine link with British society before they can apply for a student loan but set limits to how exacting a requirement this can be.⁵⁴ O’Brien remarks in this regard that

[t]he ‘real link’ case law allows Member States to attach a potentially indirectly discriminatory precondition to social assistance type benefits. It thus combines an ideal principle (equal treatment) with unprepossessing pragmatism (discrimination-lite), avoiding attracting infamous ‘benefit tourists’.⁵⁵

A real link approach, and the implied tolerance of ‘discrimination-lite’ that comes with it, is also enshrined in the Citizens directive, where the duration of a stay in the host state works as a proxy for genuine links: it is presumed that such links do not exist during the first 3 months, that they develop over the course of a longer stay up to 5 years and are strong after 5 years, when the mobile citizen can no longer be discriminated against, not even ‘lightly’.

This might prima facie seem to suggest that the genuine link requirements in the alternative Centros would not have been at odds with the case law concerning the free movement of persons, where a genuine link requirement has often been validated. However, an important difference in the configuration of cases must be noted. In all of the benefit cases, the genuine link at issue was between the migrant citizen and the host state (e.g. in Bidar, the issue was whether or not Dany Bidar had a genuine link with the UK, not with his native France). In the Centros context, the genuine link that was missing was with the UK, which, was the home state of Centros Ltd. Transposed to the context of natural persons, the genuine link requirement contemplated in the alternative Centros is similar to the one which the Spanish authorities sought to impose in Micheletti.⁵⁶ Mr Micheletti was a dual Argentinian and Italian citizen who, relying on his Italian citizenship, claimed a right to establish himself in Spain. The Spanish authorities had denied this on the grounds that his Italian nationality was not effective (indeed, Mr Micheletti had always lived in Argentina and did not appear to have any actual ties with Italy). The Court rejected the Spanish approach in terms devoid of any ambiguity.⁵⁷ Spain could not meddle with the conditions set by Italy for granting Italian nationality. It had to take Italian nationality as it comes and could not require genuine links that Italian law does not itself require. The approach followed concerning nationality in this case mirrors that of Centros regarding companies: Danish law cannot meddle with the conditions for incorporation set out by UK law. It must take a UK limited company as it comes and cannot add to the conditions set by UK company law.

In other words, there seems to be consistency across freedoms of movement in the approach regarding genuine link requirements: while Member States may be allowed to make the benefit of full non-discrimination guaranteed by EU law conditional upon a genuine link, it is always with their own territory.⁵⁸ A Member State, it seems, cannot require a genuine link with the territory of another state. It is true that exactly such a requirement was accepted by the Court in Akrich,⁵⁹ a citizenship case where the Court accepted that the UK could make the immigration of a third country national, who was a family member of a mobile citizen, conditional upon a lawful stay in another Member State (Ireland in this case). But the case was quickly overturned in Metock.⁶⁰

Centros therefore appears to be in line with citizenship cases in that the Court seems to reject a genuine link requirement when the link is with another Member State. But, as the previous sub-section shows, it is also possible to find internal market cases that are consonant with the alternative Centros. The next section seeks to make sense of this ambivalence.

4 Making Sense of the Case Law Jungle: Centros at the Crossroads

It should first be said that the tensions in the case law around how much regulatory space Member States have when it comes to neutralising the strategic use of free movement rules and imposing their rules on their territory are still very much alive. Twenty years after Centros, Polbud illustrates this in the same context of corporate mobility. The case raised the issue of whether freedom of establishment covered the strategy of a Polish company which wanted to re-incorporate in Luxembourg (and faced regulatory hindrance to its exit from Poland). Like Centros, the case was one of regulatory mobility: the company sought to have its legal seat in Luxembourg but to retain all of its activity in Poland. Although this case came two decades after Centros, both the Austrian government and the Advocate General argued that a genuine economic activity in the host state (Luxembourg) was required in order for the freedom of establishment to apply. AG Kokott wrote in her opinion that

[i]f […] Polbud seeks only to change the company law applicable to it, the freedom of establishment is not relevant. For, although that freedom gives economic operators in the European Union the right to choose the location of their economic activity, it does not give them the right to choose the law applicable to them. Consequently, a cross-border conversion is not caught by the freedom of establishment where it is an end in itself, but only where it is accompanied by actual establishment.⁶¹

The Grand Chamber did not follow AG Kokott and the judgment in Polbud is in line with Centros. The enduring tensions between this friendly attitude to corporate mobility and the cases where the Court has taken a friendlier stance towards host state regulation have been analysed in greater depth by Saydé. In a remarkable book, he shows how the issue of abuse ‘straddles three major fault lines of EU Law’, which helps to explain the inconsistencies and enduring tensions in the case law.⁶² The first fault line is not unique to EU law and opposes legal certainty and legal congruence: a doctrine of abuse necessarily jeopardises legal certainty as it leads to calling into question formally valid legal arrangements and does so in order to restore legal congruence, that is a good fit between the real situation and its legal cloak. The second fault line is specific to EU law and opposes two visions of European integration. The first vision promotes competition among firms and, in the name of a level playing field, aims to neutralise regulatory competition. It does so chiefly through harmonisation and, when necessary, through an explicit or implicit doctrine of abuse of law. The second vision, on the contrary, considers regulatory competition to be inherent in an internal market. Logically, this leads to denying any room for a doctrine of abuse. The third fault line that Saydé identifies is between two constitutional orientations: one in which the fear of private power is greater, and which is favourable to a doctrine of abuse, and the other which is characterised by a greater fear of public power, which is hostile to any doctrine of abuse.⁶³

This analysis goes to the very heart of the issue and helps us to understand that the real question is not whether the case law is consistent across freedoms of movement. The fault lines, to use Saydé’s terminology, do not run between the free movement of corporations and the free movement of natural persons. They run much deeper than this. The analytical grid that best explains the tensions in the case law cuts across legal categories and extends to fundamental differences in political orientation and value judgments.⁶⁴

Of the three fault lines identified by Saydé, the most important one in relation to Centros is the tension between regulatory neutrality and regulatory competition because it captures why Centros came as a surprise and why it is still divisive. In relation to company law, regulatory neutrality was initially consensual. Community law, in other words, was thought to promote an internal market where firms compete on a level playing field and national law was not to distort competition. Since regulatory diversity by nature will distort competition in some way, by offering a more attractive regime in some Member States than in others, the best way to achieve neutrality is harmonisation. Indeed, there were ambitious plans for an almost complete harmonization of company laws.⁶⁵ When the harmonisation project stalled, it was consensual that some obstacles to the freedom of establishment were acceptable. To compensate for the lack of harmonisation, it was thought to be legitimate for Member States to disarm regulatory competition—which had not been harmonised away but still stood in the way of fair competition among firms. In particular, Member States continued to apply the real seat theory⁶⁶ since ‘the Community [could] not tolerate the establishment of a Delaware in its territory’.⁶⁷Centros blew away this consensus in favour of regulatory neutrality. For the first time, very explicitly, the Court embraced the vision that the internal market is about regulatory competition. This stance is not exactly unique to company law and it also surfaces in the case law on nationality (Chen, Micheletti),⁶⁸ but nowhere is it as explicit or as radical.

5 Concluding Remarks and Some Thoughts for Further Research

Centros is not an oddity in internal market law because there are many other cases in which the Court has reaffirmed that Member States may avail themselves of the necessity to combat abuse in order to justify measures which restrict free movement but has denied that the conduct at hand amounts to an abuse or that the anti-abuse measures are proportionate. It is not uncommon, therefore, that the Court seems to be paying lip service to a doctrine of abuse in order to soothe Member States and alleviate their concerns about the deregulatory effects of freedoms of movement, while allowing regulatory diversity to feed regulatory competition. What is more unusual in Centros is that the Court was uncommonly forthright about what it was doing. It was not only saying that there was no abuse. It was saying that playing one law against another is the essence of free movement. This is what caused commentators’ surprise and, as the case may be, joy or concern. It was always understood that the internal market was there to guarantee a level playing field in the economic competition between undertakings. It was not always understood that the internal market meant setting states in regulatory competition against one another. Centros made this dimension truly explicit for the first time.

Other than that, Centros fits well in the ebb and flow of internal market law. Based on a constant analytical framework (do the state measures hinder free movement? If so, are they justified?), the Court regulates the allocation of powers between the Union and Member States as well as between the home state and the host state without much—or indeed any—need for a formal doctrine of abuse. Whether the Court takes a more or less friendly stance towards host states depends on the state of harmonization and, in the absence of harmonization, on the quality of the justifications as well as the Court’s own value judgments regarding the merits of regulatory competition in different sectors. The Court encouraged regulatory competition in company law as in no other field of law. At the other end of the spectrum, in tax law, it offered the most explicit protection to host state powers.

What the preceding discussion has added to the already rich literature on these questions is an observation concerning the genuine link requirement. While Member States are often at liberty to use a ‘centre of gravity’ technique and to impose a (proportional) requirement of a genuine link with their territory (as with student loans and grants, for example), there is no example (after Metock) where it has been accepted that a Member State requests a genuine link with the territory of another Member State. Besides Centros, Micheletti is another case in which the issue of whether a Member State may require a genuine link with another Member State arose. In neither case, was any interference tolerated with the home state’s policy on the nationality of either persons or corporations. It is always for the home state to define who is a national (Michelletti, Chen) or what it takes to incorporate in its legal order and the host state cannot call that choice into question. This possibly constitutes the strongest objection to the alternative Centros examined in Sect. 2. It may also help to explain why EU law, acting as an umpire in the allocation of regulatory competences between Member States, sometimes protects the regulatory competence of home states, thus sparking regulatory competition, while in other cases it protects the competence of host states (not only in matters of tax, but also, for example, concerning professional training or gambling).

However, this is a long way from a complete explanation of the allocation of regulatory powers between the home state and the host state in all cases when this allocation is not already made clear by the Treaty or secondary legislation. In this regard, my suggestion for a further exploration is to look at the issue from the point of view of costs and to investigate whether thinking of these costs through a behavioural lens in terms of losses helps to make sense of the case law. One issue worth investigating in this regard is whether the perception of costs explains the normative attitude to regulatory competition.

Three attributes of costs may be relevant: their magnitude, salience and attribution. Any regulatory choice, particularly in the economic sphere, entails costs and potential gains. Increasingly, impact assessments and other forms of cost–benefit analysis seek to measure net gains or costs. The magnitude of the costs depends on how people react to the rule. For example, the cost of student support will depend on how many out-of-state students are attracted (which is in turn impacted by EU law facilitating free movement). If German businesses react to Centros by shopping for more advantageous company law regimes, and, instead of incorporating in Germany, go to the UK, there will be lost revenue for German notaries and its magnitude will depend on how many companies choose the UK Limited company over the German GmbH.⁶⁹

Irrespective of their magnitude, some costs are more salient than others. For example, when a state has to pay out maintenance grants to out-of-state students, unemployment benefit to migrant workers or social benefits to migrant citizens and their family members, such costs are easy to see and register vividly with decision makers, with the public and, in all probability, also with the Court of Justice. By contrast, the cost of higher unemployment in a particular sector in 2 years are much less salient and are likely to be ignored or have a lesser impact on decision making. One hypothesis that would be worth testing is whether salient costs matter more than non-salient costs. I suspect that they do.

Another possibly relevant dimension of costs relates to what one might call attribution: how people attribute (rightly or wrongly) the costs that the state has to bear is likely to impact their judgement as to whether such costs should be incurred or not. Some costs may be the direct and inevitable consequence of the state’s own regulatory choices (such as enforcement costs), others may be triggered by a change in the law of another state (which has just made its company law more attractive). Whether costs are attributed to the state’s sovereign decision, to a foreign state, to ‘migrants’ or to ‘Europe’ might plausibly impact the perception of the desirability of regulatory competition but this is probably more difficult to study than the salience and magnitude of costs. To begin with, it would be interesting to assess how much explanatory power these two dimensions have and to test whether the Court tends to protect the regulatory prerogatives of states more when the costs imposed on them by the operation of mobility are large and/or salient.

Quite possibly, the costs induced by the mobility of natural persons (in the form of various social benefits) are generally more salient to states than the costs induced by corporate mobility. Could this explain in part why EU law seems to respond differently to various instances of Member States trying to limit their exposure to the costs of mobility? Are ‘genuine link’ requirements more readily accepted when the potential costs of not having them would be salient or large? Despite the well-established principle that purely economic justifications are not valid,⁷⁰ is there an underlying economic or psychological logic that could help explain when internal market law is more or less tolerant of Member States’ efforts to limit the extent of their cost exposure associated with free movement?

One reason to believe that such a logic may be at work is linked to loss aversion. Kahneman and Tversky, two psychologists whose joint work earned Kahneman the Nobel Prize in economics,⁷¹ showed that most people are loss averse (rather than risk averse) and that losses loom much larger than gains (about 2½ times larger in their experiments).⁷² Most people, for example, will not risk tossing a coin if they can lose € 100 or gain € 100. It is necessary to raise the potential gain to € 200 or € 250 to get most people to agree to play. The question this raises in the context of Centros is whether the Court might be sensitive to states’ loss aversion and embraces regulatory competition only when the costs it imposes on states are (perceived to be) not too large or not too salient.