The two reference cases considered here concern the compatibility with EU law of industrial disputes and collective actions against EU companies exercising their free movement rights. The Swedish case, under a reference of the Arbetsdomstolen (Swedish Labour Court) of 15 September 2005 in litigation between Laval un Partneri Ltd (hereafter ‘Laval’) v. Svenska Byggnadsarbetareförbundet, Avdelning 1 of the Svenska Byggnadsarbetareförbundet, Svenska Elektrikerförbundet (in the material that follows, ‘Bygnadds’) *2 concerns the question of whether industrial action of Swedish labour unions against a Latvian company that wanted to perform a work contract under Swedish procurement rules through the use of posted Latvian workers falls under the ‘freedom to provide service’ rules of article 49 EC and, if this is the case, whether this action can be justified either under the posted workers directive, 96/71/EEC *3 , or under a specific Swedish law exempting labour unions from liability in taking action against foreign-based companies (the so-called Lex Britannia; see section 5.1 below).
The reference from the English Court of Appeal of 23 November 2005 in the litigation between (1) the International Transport Workers’ Federation (ITWF) and (2) the Finnish Seamen’s Union (FSU) versus (1) Viking Line ABP and (2) OÜ Viking Line Eesti (referred to below as the Viking case *4 ) concerns the question of how far labour unions can take social action against a reflagging of a shipping company from a ‘high-wage’ country (Finland) to a ‘low-wage country’ (Estonia); the ECJ was asked to decide also on the applicability of article 43 EC as well as Regulation 4055/86 *5 in the litigation, including possible justifications. In more general terms, the case concerns the so-called FOC (flag of convenience) policy of the ITWF aimed at eliminating FOCs by establishing a genuine link between the flag of the ship and the nationality of the owner, and by protecting and enhancing the conditions of seafarers serving on FOC ships.
Both reference cases concern delicate matters of how to balance social policy objectives with economic freedoms that became apparent in the EU following the accession of ten (and now 12) new member countries. Most of these new member countries still have much lower wages, which give them a competitive advantage in the internal market but may easily be challenged (and indeed have been challenged by social action in the host countries) as provoking ‘social dumping’ on the more elaborate wage policies of old — in particular, Nordic — member countries. Both the Laval and (perhaps to a somewhat lesser extent) the Viking case have aroused strong political reactions in the Member States concerned, although these will not be addressed here. The aim of this article is more modest: it is intended to offer a legal analysis under existing EU law concerning how to solve these conflicts. It takes, as a starting point, the existing case law of the ECJ, which, however, has not yet really resolved the new types of conflict that arose in the Laval and Viking cases. Therefore, in deciding on the references, the Court must provide a truly constitutional answer concerning how to settle the existing — and possible future — conflicts between social structures in Member States that still remain within their own area of competence and the dynamics of EC law seemingly favouring the liberal spirit of free movement to the detriment of the social arrangements of Nordic countries with a strong welfarist tradition in particular, based upon the central role played by autonomous labour unions enjoying far-reaching action rights.
The analysis starts from the premise that the particular type of conflict that is before the ECJ in the Laval and Viking cases has not been regulated by the rather elaborate transition arrangements in the accession treaties. It must be remembered also that the actions are directed not against a Member State, as is the usual setting in free movement cases, but against labour unions, which are governed by private, not public law, and which enjoy, in the traditions of all Member States — whether old or new — a substantial amount of autonomy guaranteed by national and European constitutional provisions. Therefore, it must be analysed first how far this constitutional autonomy extends within the system of the EC free movement rules (see section 2). Only then should the question be answered of whether the relevant free movement provisions — namely, those concerning services (article 49 EC), regarding Laval, and on establishment (article 43/48 EC) in the Viking case — can be applied to social action by labour unions restricting free movement of posted workers with respect to reflagging by shipping companies (see sections 3 and 4). Should the answer be positive, one has to look for possible justifications, which will be different for Laval on account of the existence of the posted workers directive, 96/71 (see section 5), which is not applicable in the Viking case (see section 6). Finally, the recent opinions of Advocate General Mengozzi and Poiares Maduro of 23 May 2007 will be mentioned (in section 7).
2.1. Possible exemptions of labour unions from article 49/43 EC
It could be — and has been — argued that labour unions taking social action in industrial disputes are exempted from the application of Community law in general and particularly from article 49/43 EC, which is directed against States only. Collective action by labour unions is, according to such an opinion, meant not to restrict freedom to provide services but to ensure adequate conditions of work and pay. Under article 137 (5) EC, the EU does not have jurisdiction in matters of strike, lock-out, and pay. In more general terms, it could be (and, again, has been) argued that Title XI on social policy leaves this area to Member States and allows only very limited intervention on the part of the EU. This seems to imply the non‑applicability of the fundamental freedoms to industrial actions, arguments put forward in particular by the labour unions in both cases and by the supporting Swedish and Finnish governments, as well as, indeed, by most governments from ‘old member countries’ that had submitted observations to the Court.
In the face of such an argument, it should be clarified that the exclusion of Community legislation in the field of industrial action, on the other hand, does not preclude the presumption that the general principles of Community law always must be respected, as the Court has frequently held with reference to other prerogatives of Member States — e.g., with regard to direct taxation *6 , health provisions *7 , social security *8 , higher education stipends and loans *9 , war pensions *10 , or the property regime. *11 There is no reason to exclude a priori social policy matters from the application of free movement principles. Article 137 (5) only excludes Community legislation in the area of strikes and lock-outs, not the effects of primary law on industrial action.
One could refer also to the limited application of Community competition rules to collective bargaining and industrial action, which should be taken over by analogy to the EU freedoms. With regard to competition law, it is a matter of debate, under the influence of US/American antitrust law, whether there is an inherent ‘non-statutory exemption’ for collective action in industrial relations. *12 The ECJ, in its Albany judgment, took a somewhat more restrictive view:
Under an interpretation of the Treaty as a whole which is both effective and consistent [...] agreements concluded in the context of collective negotiation between management and labour in pursuit of such [social policy] objectives must, by virtue of their nature and purpose, be regarded as falling outside the scope of Art. 81. *13
This argument must, however, be limited to competition law where business and labour organisations are negotiating for a pension fund for the employees of a particular sector, as in the Albany case. The case did not concern industrial action as such and therefore is not a precedent for the limitation of actions by labour unions with regard to fundamental freedoms. Where collective action by labour unions confronts individual businesses from other member countries by rendering their market access or business restructuring impossible or more difficult, there is no ‘non-statutory exemption’ from the application of the EU free movement rules.
An important argument in proposing to exempt collective action by labour unions from the applicability of the fundamental freedoms could be based on article 11 of the European Human Rights Convention (EHRC). Since the EU, according to its article 6 (2), has to ‘respect’ fundamental rights, and since the ECJ, in its case law concerning fundamental freedoms, refers to judgments of the European Court of Human Rights (ECtHR) *14 , it is helpful for the Laval and Viking cases to look at precedents from the Strasbourg Court. The ECJ is, of course, not formally bound by the case law of the ECtHR, but it will usually pay tribute to it if questions of fundamental rights in the sphere of application of EU law — in particular, with regard to the exercise of fundamental freedoms — are at stake. This is all the more true in areas where no consistent case law of the ECJ exists, as with regard to the conflict between industrial action and fundamental freedoms.
Article 11 (1) ECHR guarantees the freedom of association, subject to limitations spelt out in paragraph 2. This right can be invoked by labour unions in collective bargaining and action. It encompasses a positive element allowing social action including strikes for the improvement of working conditions, and therefore it may conflict with the fundamental freedoms of the EC if such actions necessarily restrict free movement of goods or services.
On the other hand, this ‘positive right of association’ that can be invoked by Bygnadds in the Laval case and by ITWF / the Finnish Seamen’s Union in Viking is not without limits. It must respect the rights of other individuals to freely associate or not to associate. This problem was raised before the European Court of Human Rights in the case Gustavsson v. Sweden . In its judgment of 25 April 1996 *15 , the Court faced the question of the extent to which industrial action against an entrepreneur who did not want to join a collective bargaining agreement was permitted or, conversely, limited by article 11 of the EHRC.
This judgment recognised that ‘freedom of association’ also has a ‘negative side’, which Sudre calls droit d’assocation négatif *16 — namely, the right of an employer not to be forced into a collective bargaining agreement if he does not belong to the relevant trade association. This right is enjoyed also by Laval and Viking in offering services in Sweden respectively Finland/Estonia. Its exercise is made impossible or severely restricted by the industrial action of labour unions. Article 11 of the ECHR therefore does not exclude the application of 49 EC; rather, on the contrary, it must be read alongside article 11, with the aim of protecting both the free movement and the negative aspect of the right to association.
This case law was confirmed by the recent Sorensen v. Rasmussen judgment handed down against Denmark *17 , where it was clearly stated that the requirement to join a certain trade union violates article 11 ECHR, as such a requirement has an effect on the very freedom of assembly and association. Therefore, the state must avoid measures violating this ‘negative right of association’, even though it enjoys a certain margin of appreciation. Obviously, the primary addressees of such an obligation are the unions themselves, which must avoid infringement of the ‘negative right to association’. Their actions will have to be balanced against the two components of article 11 ECHR: their ‘positive right’ to association and the ‘negative right’ of social partners who do not want to adhere to a system of collective bargaining. In the EC context, this balance must be accomplished within the ambit of the free movement provisions of the treaty, which will be analysed in sections 3 and 4, subject to justification under the proportionality principle outlined in sections 5 and 6.
In its Werhof judgment *18 , the Court derived from article 11 of the EHRC as interpreted by the ECtHR the principle that “[f]reedom of association [...] includes the right not to join an association or union”. This ‘negative right to association’ must be respected in interpreting secondary Community law, e.g., in the case here before it: the directives on transfer of undertakings. The Court expressly referred to the Gustavsson judgment of the ECtHR, thus indicating that it takes the same view on the limitations of the right to collective action. This demonstrates clearly the concerns of the ECJ that social action rights of labour unions are not ‘outside’ EC law (whether primary or — as in the Werhof case — secondary law) but must be balanced, in particular, against existing free movement rights. This will be analysed in later sections of the paper with regard to the freedom to provide services (section 3) and the right to establishment (section 4).
According to article 6 (2) EU, the Union will also respect fundamental rights, as they are common to the constitutional traditions of the Member States. There is agreement that one of these traditions concerns the rights of social partners — labour unions and business organisations — to take social action to defend their legitimate interests in industrial disputes, but this right is not without limits. The right has been recognised in the Charter of Fundamental Rights in the EU — namely, in its article 28 with the following words (emphasis added):
Workers and employers, or their representative organisations, have, in accordance with Community law and national laws and practices, the right to negotiate and conclude collective agreements at the appropriate levels and, in cases of conflict of interest, to take collective action to defend their interests, including strike actions.
In particular, the Nordic states defend their specific social model, which is based on considerable autonomy granted to social partners by their constitutions without the state necessarily interfering in social action. *19 It is feared that this autonomy of the ‘Scandinavian social model’ may be endangered if EC law interferes in its workings by imposing its specific rules on free movement or — as had been discussed in Albany *20 — on imposing competition law on social action.
In the context of this discussion, it suffices to remember that even in the Scandinavian model of industrial relations there are certain inherent limits to social action, which are spelt out by law. The most important and common one is the so-called ‘peace obligation’ (or Friedenspflicht), which forbids industrial actions once a valid collective agreement has been concluded between social partners. Of course, the extent of this Friedenspflicht and the consequences of its breach may be very different, but they are in any case determined by law, not by the social partners themselves, as can be shown very clearly in examination of the Swedish case of the Lex Britannia, which will be discussed below (in section 5.1). In Finland , article 13 of the Constitution protects freedom of association of social partners, including the right to strike, which is subject to three limitations, namely where the right to strike is outlawed by a Finnish statute, where the strike is contra bonos mores, or where the strike is in breach of EC law directly applicable between the parties. *21
This is exactly what is stated in article 28 of the Charter of Fundamental Rights, whose indirect legal importance recently was recognised by the Court. *22
The text of article 28 makes it very clear that the right to industrial actions exists only within the limits of the law, including EC law. Of course, these limitations must respect the principle of proportionality, as is spelt out in article 52 (1) of the Charter. But it cannot be said that industrial action as recognised by MemberState law is completely outside the scope of Community law. This has also indirectly been recognised by the Court in Albany, where it defined, on the one hand, the existence of an exemption of social partners from the competition rules in order to attain legitimate social policy objectives but implicitly rejected any extension of this exemption beyond matters of social policy, on the other. It is therefore a matter of balancing the different objectives of the EC Treaty — free movement — against social action (namely, adequate social protection of workers as mentioned in article 136 EC). Exactly this balancing is now before the Court in the Laval and Viking cases. Any one‑sided approach to resolving this question must be rejected — either by giving absolute precedence to social rights or by formally insisting on the supremacy of free movement without taking account of legitimate social policy objectives pursued by the Member States, even when delegated to social partners as in Sweden and Finland . The following sections attempt such a balancing approach.
3.1. Posting of workers by Laval in Sweden as exercise of the freedom to provide services
With regard to applicable Community law ever since Rush Portuguesa *23 , it is without doubt that the posting of workers of a company established in one EU country is a (cross-border) service to which article 49 EC is applicable. The workers employed by Laval are not seeking access to the Swedish labour market but will be removed once the construction work as contracted is finished. *24 In principle, they remain under Latvian jurisdiction. Therefore, the provisions concerning free movement of workers (article 39) and non-discrimination (article 12 EC) can be disregarded in this context.
It should be mentioned also that Sweden did not make use of the possibility of invoking a transitory regime against the posting of workers, as had been conceded to Germany and Austria during the accession negotiations with the new member countries, including Latvia . *25
As the case shows, the boycott by Bygnadds is the strongest form of restriction; indeed, it made impossible the rendering of services by Laval in Sweden and caused great harm both to Laval and to the Latvian workers it had posted while relying on its freedom to provide services. Article 49 EC is directed both against ‘discriminations’ and against ‘restrictions’. *26 For a preliminary examination of the legality of the boycott under EU law, it is sufficient that there exist a restriction that is forbidden unless it can be justified. The question of discrimination will be discussed in considering the possible justifications (in section 5.1.3).
Under EU law, this action constitutes the very negation of the freedom to provide services. In its effects, the restriction forced upon Laval by the action of Bygnadds is to some extent similar to the re-incorporation requirement imposed by German law on foreign companies. In its Überseering judgment *27 , the Court wrote with regard to establishment, an argument that can be employed also with regard to freedom to provide services: “The requirement of re-incorporation of the same company is therefore tantamount to outright negation of the freedom of establishment.” This restriction can lead to a double set of consequences under the case law of the Court: an obligation of the (Swedish) state under article 10 EC to ensure that freedom to provide services is guaranteed, not to be discussed here, and an effect against the Swedish labour unions themselves under the ‘horizontal direct effect’ theory of article 49 EC.
The main legal question of the case concerns the effects of article 49 EC against the action of Swedish labour unions. Case law of the ECJ is evolving; the answer is not yet clear with regard to the position of labour unions in industrial action infringing free movement.
The traditional approach to the freedoms enshrined in the EC Treaty has been concerned exclusively with regard to state action. This seems to be suggested by their very place within the EC Treaty, in article 3 (1) c) and 14 (2), concerning the establishment of an internal market without frontiers among the Member States. Restrictions of market access by private persons usually fall within the net of the provisions on competition, per article 81/82 EC. This seems to exclude the applicability of article 49 and the other Community freedoms to privately imposed restrictions.
Such a narrow interpretation, however, is not required, from the very wording and system of the treaty. *28 Article 49 (1) EC itself takes a somewhat broader view, as it aims at abolition of restrictions on the freedom to provide services, whatever their origin. The Court, in a series of cases, therefore extended the applicability of article 49 (and other freedoms — namely, those addressed in article 39 and article 43) to privately imposed restrictions. This case law started with Walrave *29 already in 1974, concerning restrictions on free movement imposed by bylaws of sporting associations. This case law, which takes a functional rather than formal approach to interpreting the fundamental freedoms, has been continued in the well-known Bosman case. *30 This case was concerned with free movement of workers, even though the Court also referred to services. This precedent was confirmed and justified in Angonese. *31 In the later Wouters case *32 , the Court summarised and confirmed its practice law with regard to collective regulation by private entities.
In the case at hand, the boycott of Bygnadds imposed on Laval was intended to achieve a ‘regulation’, the conclusion of a collective bargaining agreement or a supplement according to Swedish law. This would have resulted in Laval having to pay Swedish wages and benefits, thereby loosing the competitive advantage that allowed it in the first instance to win the Vaxholm contract. At the same time, it would have set aside the agreement that Laval had concluded with Latvian Construction Workers’ Association (CWA) in summer 2004.
As a preliminary result of this discussion, it can be clearly stated that the boycott by Bygnadds violated Laval’s right to freely provide services. Labour union actions are not exempted from prima facie application of the free movement rules. Since article 49 EC is applicable in this context, other EC provisions need not be discussed in detail. Bygnadds may be liable for breach of Community law if its action cannot be justified, but this will not be discussed here.
The right of establishment set forth in article 43/48 EC can also be invoked horizontally against collective actions as defined above. The arguments presented there need not be repeated insofar as the Laval and Viking cases concern similar legal questions. *33
The origin of the social conflict, however, is somewhat different because it concerns the general FOC policy of the ITWF (see section 1 above). When in 2003 the FSU was confronted with the request of Viking Finland to reflag its ferry Rosella, which serves the Helsinki–Tallinn line, in order to avoid losses due to higher wages paid under Finnish law and be allowed to pay wages according to lower (Estonian) standards, it contacted the ITWF, which sent a circular to all affiliated organisations asking them to refrain from negotiating with Viking. Since the collective bargaining agreement with Viking concerning the manning of Rosella ended in November 2003, the FSU threatened social action against the intended reflagging. Viking asked for an injunction in the Finnish labour court, which was refused. The mediation under Finnish law ended with Viking giving in to the demands of the FSU. It concluded a new collective agreement ending on 28 February 2005, which renewed the old agreement; it had to give up its plans to reflag the ship. Since the Rosella continued to make a loss and since the ITWF circular remained in force, Viking decided to bring proceedings before the English High Court in August 2004 seeking declaratory and injunctive relief, which required withdrawal of the ITWF circular and requiring the FSU not to interfere with Viking’s free movement rights in relation to reflagging of the Rosella. The High Court granted the injunction in favour of Viking in June 2005. This judgment was appealed by ITWF before the Court of Appeal, which quashed the judgment for an injunction and decided to make the above-mentioned reference.
The specific problem caused by the threatened social action in the Viking case concerned restrictions on a company wanting to partly move from one jurisdiction to another via the intended reflagging. According to question 6 of the referring Court of Appeal, this is done by the
parent company established in Member State A (Finland) […] intend[ing] to undertake an act of establishment by reflagging a vessel to Member State B (Estonia) to be operated by an existing wholly owned subsidiary in Member State B which is subject to the direction and control of the parent company.
In my opinion, it must be argued that the social action taken against the reflagging by Viking must be regarded as a restriction of its freedom of establishment, which would be illegal unless it can be justified by — proportionate — social policy reasons (see section 6).
The main problems of the case concern with the question of justification, as the Swedish Labour Court seems to imply itself. The examination must involve certain steps that are suggested by the questions of the referring court, namely:
– the Swedish Lex Britannia (1)
– the importance of directive 96/71 (2)
– general principles of Community law (3)
The Swedish Lex Britannia amended the Medbestämmandelagen (or MBL, law on workers’ participation in decisions) in 1991 *34 and allows Swedish labour unions to commence industrial action against an undertaking that has not (yet) concluded a bargaining agreement with a representative labour union of its employees. At the same time, there is a Friedenspflicht — that is, a duty to abstain from industrial action — only in those cases where a binding collective agreement already exists. The 1991 amendment limited this Friedenspflicht to those agreements to which Swedish law ‘directly’ applies, thus excluding collective bargaining agreements concluded with non-Swedish labour unions. Industrial action therefore is not prohibited when a foreign employer carries out temporary activities in Sweden and an overall assessment of the situation leads to the conclusion that the connection to Sweden is so weak that the MBL cannot directly apply to industrial relations. The 1991 amendment was designed to combat ‘wage dumping’ by foreign service providers made possible by the prior case law of the Arbetsdomstolen, whereby Swedish labour unions had to recognise foreign collective bargaining agreements in the sense of the Friedenspflicht even if the wages negotiated were below Swedish standards. According to the travaux préparatoires, the Lex Britannia is intended to allow trade unions to act to ensure that all employers active in the Swedish market pay salaries, and grant other conditions of employment, in line with those usual in Sweden and that they create conditions of fair competition on an equal basis between Swedish undertakings and entrepreneurs from other countries. However, its potential discriminatory effects against undertakings established in other EU Member States that want to post workers in Sweden must be examined.
The Lex Britannia does not contain a specific exemption concerning industrial action against undertakings established in an EU Member State. This ‘omission’ by the Swedish legislators was criticised by several Swedish experts when Sweden acceded to the EU *35 , but it has not been taken up by the Swedish legislators in the ratification proceedings of the treaties with the new Member States. There is no exclusion in Swedish law concerning the continued application of the Lex Britannia vis-à-vis undertakings from new member countries. The Swedish legislative system wants to apply the Lex Britannia also against service providers established in the EU. Simultaneously, the general provisions of the freedom to provide services apply in full, as has been demonstrated above (see section 3) against industrial action. Only when there is specific justification for a restriction to provision of services may the enabling provisions of the Lex Britannia continue to be applicable and be invoked to justify industrial action on the part of Bygnadds.
The case law of the ECJ with regard to the restrictions allowed to article 49 EC is relatively well developed and quite clear in its main principles:
– restrictions may be justified to protect posted workers;
– even if allowed, these restrictions must be proportionate and non‑discriminatory;
– restrictions are not allowed to close the national construction market and to shield national businesses from competition by service providers established in the EU.
Already in the case Rush Portuguesa *36 the Court said (paragraph 18):
Community law does not preclude Member States from extending their legislation, or collective labour agreements entered into by both sides of industry, to any person who is employed, even temporarily, with their territory, no matter in which country the employer is established; nor does Community law prohibit Member States from enforcing those rules by appropriate means.
Obviously, this exemption does not apply in the Laval case. The Lex Britannia is, as I have argued, enabling legislation, not protective legislation as in the Rush Portuguesa case.
Later cases impose certain limits on legislation protective of workers. In Finalarte et al v. Urlaubs- und Lohnausgleichskasse der Bauwirtschaft *37 , the Court insisted that it may serve the protection of workers but “cannot be justified by economic aims, such as the protection of national businesses”. The yardstick for legitimate restrictions of the freedom to provide services is always the protection of posted workers, not the prevention of competition in the construction market. *38
The neutrality of the Swedish state is in reality a delegation of power to labour unions to restrict the EU freedoms, if the undertaking providing services from another EU country has not complied with Swedish labour standards. Labour unions, of course, are not forced to take such action; they are only enabled to do so by the privilege of being exempted from civil liability. But, as practice shows, they usually take action if a home business (as in the Gustavsson case) or an EU undertaking (as in Laval) does not comply with Swedish labour standards. Since the Swedish state has given them power to restrict the fundamental freedoms, they can therefore be held liable for violations of Community law and cannot escape their responsibility by relying on the Lex Britannia, which by its very wording allows them to impose such restrictions (in its sub. 3).
Since the Lex Britannia is an enabling law, its discriminatory elements can be seen only once it is put into action. The Laval case gives striking evidence in this direction. Bygnadds used the exemption from Friedenspflicht allowed by the Lex Britannia to initiate industrial actions against Laval that in the end made impossible the rendering of cross‑border services in Sweden . Laval sought an injunction against this action in order to be able to continue providing its services as agreed under the construction contract with the city of Vaxholm, which was refused by the first judgment of the Arbetsdomstolen. The combined functioning of the Lex Britannia as an enabling law for Bygnadds and as an exemption from liability prevented the provision of services. The Lex Britannia therefore is the relevant and decisive cause for an unjustified restriction of article 49 EC.
Compared to the situation of a Swedish undertaking, that of Laval clearly involved being treated in a discriminatory way. Under the existing practice of the Lex Britannia, it could not invoke its collective bargaining agreement with LACW as an element of the Friedenspflicht that would not allow industrial action by Bygnadds and could therefore be prevented by an injunction before the Swedish labour court.
It cannot be argued that Bygnadds merely wanted Laval to offer its services on the same terms as Swedish construction companies, and that they therefore were treated on equal terms. Such an argument disregards the difference between Swedish and Latvian undertakings in providing services. By forcing upon them ‘equal conditions’, in reality existing differences are levelled out, downward. This is the opposite of non‑discriminatory treatment: Laval is deprived of its competitive advantage and to that extent discriminated against. Bygnadds is using the instruments of the Lex Britannia to impose restrictions on the freedom to provide services on economic grounds, which cannot be justified. Since the Lex Britannia does not prohibit such industrial action, this omission by the Swedish legislature clearly violated Community law; such violations cannot justify action by Bygnadds.
Directive 96/71 was adopted under internal market jurisdiction concerning provision of (cross-border) services. It therefore is an attempt to implement the basic principle of article 49 EC as it had been developed by the case law of the ECJ, particularly since Rush Portuguesa. This starting point must be kept in mind in interpreting it. Therefore, the directive should not be applied in such a way as to make cross-border provision of services via posted workers impossible or unreasonably risky or costly. The host MemberState must guarantee that posting remains possible within the framework of the directive. It should not be used on mere economic grounds in order to protect national businesses or close markets to competition.
At the same time, the directive, according to recital 5, is aimed at promoting a “climate of fair competition and measures guaranteeing respect for the rights of workers”. *39 Therefore, a “‘hard core’ of clearly defined protective rules should be observed by the provider of services notwithstanding the duration of the worker’s posting”, according to recital 14. The second objective of the directive is the protection of posted workers, not of workers of the host country. *40 These rules are contained in article 3. They imply also a conflict rule — namely, that not only the law of the home country of the service provider but also in certain respects the law of the host country is applicable (that is, the law of the country where the posted workers perform the work according to the service contract that the provider has concluded in the host country). To this extent and aim, it amends the rules of applicable law under article 6 of the Rome Convention to protect posted workers and to make possible their equal protection under the law of the host MemberState. *41
As far as the Laval case is concerned, it seems to me that only the question of ‘minimum rates of pay’ in the sense of paragraph 1, second indent, lit. c) is controversial. *42
The directive sets out a number of requirements that must be observed in order to ‘overrule’ the rates of pay that have been agreed upon by the provider with ‘his’ workers, or that are applicable according to a collective agreement concluded in the home country, as in the Laval case. The conditions that are listed in the two indented portions of paragraph 1 of article 3 — namely, minimum wages imposed either by law (or regulation) or by collective agreements and arbitration awards declared universally applicable — are not relevant here. Sweden does not have a system of state minimum pay (as in France ), nor one of declaring collective agreements universally applicable within the meaning of paragraph 8 (as in Germany ). Sweden deliberately refrained from implementing lit. c) because of its ‘social model’, which means that the state takes a neutral position with regard to industrial relations and leaves it to the social partners to find adequate pay rates in collective bargaining proceedings, which are enforced by the mechanisms available under the Medbestämmandelagen, as amended by the Lex Britannia.
Paragraph 8 of article 3 extends the effects of collective agreements with regard to pay rates that may also be imposed on posted workers under certain conditions, which so far have not yet been interpreted by the ECJ. They therefore need careful scrutiny.
(1) The first condition requires a decision of the host MemberState; namely (emphasis added)
In the absence of a system for declaring collective agreements or arbitration awards to be of universal application […], Member States, may, if they so decide, base themselves on […]
As has been correctly observed by Davies *43 , “in determining whether collectively agreed standards apply for the purposes of the directive, Member States may substitute for the test of obligation the test of applicability in fact”.
The Arbetsdomstolen, in its order for reference, explains that “the responsibility for deciding what is an acceptable standard for terms and conditions of employment has thus been transferred to the organisations in the Swedish employment market”. *44 It is not clear how this has been accomplished, particularly vis-à-vis foreign service providers who may not know the specifics of the Swedish labour market. The directive is silent as to the form and requirements of such a decision. By argumentum e contrario, it need not be a legally binding instrument. On the other hand, from the very purpose of the directive — namely, not to serve as a basis for preventing competition and at the same time to protect posted workers — it must be required that such a decision be expressly taken and be rendered transparent for the industries listed in the annex, which is applicable to the Laval case, concerned with construction work. Both the foreign service provider and the posted workers must know under what conditions they perform work in the host state, as expressly spelt out in article 4 (3) of directive 96/71. It should be remembered that a similar requirement of transparency has been developed by the ECJ in public procurement as well as with respect to state aid matters if Member States want to enforce certain justified protective objectives. *45 It is not known whether Sweden has taken such an express decision, and whether it was transparent to Laval during the tender process for the construction contract in Vaxholm.
(2) Even if there is no doubt concerning the (express and transparent) ‘decision’ of the Swedish state, as a second requirement the collective agreement must either “be applicable to all similar undertakings in the geographical area and in the profession or industry concerned” or have been “concluded by the most representative employer’s and labour organisations at national level and that apply throughout the national territory”. According to the EC-Commission, this means that the Member States can include agreements or awards that are complied with by the great majority of ‘national-level undertakings’. The key factor is the extent to which the national-level undertakings are real or potential competitors to the service provider. *46 Obviously, this is a question of fact that cannot be answered by the ECJ. However, serious doubts exist that these requirements have been fulfilled in the Laval case:
– it is not clear that the wage level agreed to in a collective agreement in the construction business in the Vaxholm (Stockholm) area amounts to the originally required 145 SEK per hour;
– even if Bygnadds has concluded a collective agreement with the relevant Swedish employers’ association as the most representative organisation, it is not evident that this agreement is applied ‘throughout the national territory’ rather than only regionally.
Bygnadds seems to take the view that article 3 (8) must be understood in such a way that it also allows taking industrial action to extend the application of collective agreements to foreign service providers. It refers to recital 12 of the directive, which reads:
Whereas Community law does not preclude Member States from applying their legislation or collective agreements entered into by employers and labour, to any person who is employed, even temporarily, within their territory, although his employer is established in another MemberState; whereas Community law does not forbid Member States to guarantee the observance of those rules by the appropriate means […]
As can be seen from the very wording of the recital, it only reiterates what has been said already in paragraphs 1 and 8 of article 3 of directive 96/71. It says nothing about corresponding rights of trade unions but speaks only of Member States. The argument that in the case of Sweden the state, by virtue of the Lex Britannia, has delegated this power to trade unions is irrelevant because the Lex Britannia itself violates Community law (see section 5.1.3 above) and therefore cannot serve to justify the action of Bygnadds to extend its collective agreements to EU service providers.
The same is true of recital 22 of the directive, to which the order for reference of the Arbetsdomstolen refers. It states that the directive is without prejudice to Member States’ legislation governing the right to take industrial action. It does not say anything — and cannot do so — about the limits of industrial action under primary Community law, as has been set out above.
(3) It should be remembered that paragraph 8 of article 3 refers to paragraph 1, including lit. c), which discusses only ‘minimum rates of pay’, not ‘normal pay’ as agreed in collective bargaining. There is some doubt that the Swedish system allows such a distinction to be made between ‘minimum’ and ‘normal’ rates of pay. Such a distinction is, however, necessary in order not to take away from foreign service providers the competitive advantage that they may possess by paying lower wages to their workforce under the jurisdiction of their home country, while they may still incur higher costs because of offering work in a different country, where they may need to compensate for costs of transportation, shelter, and living of the posted workers. Application of the normal wage rates in the host country may be disproportionate in special circumstances, as the Court spelt out in Portugaia Construções. *47
(4) Finally, a fourth and decisive requirement, that of ‘equality of treatment’, must be guaranteed. This ‘equality’ principle is violated when the collective agreement allows or at least tolerates an undercutting of the agreed pay rates while such practice is not tolerated vis-à-vis foreign service providers. This principle had been developed by the ECJ in Portugaia Construções with regard to collective agreements that have been declared universally applicable, as in Germany . *48 A fortiori it must also be applied to collective agreements that are de facto universally applicable. *49
Even if one takes the view (which I do not share) that the requirements of article 3 (8) are met in the Laval case, and that therefore the posted Latvian workers should have been paid at hourly wage rates of around 145 SEK, a refusal to do so by Laval does not automatically mean that industrial action by Bygnadds would be justified for imposing a collective bargaining agreement upon Laval. Article 5 (2) of the directive requires that Member States “in particular ensure that adequate procedures are available to workers and/or their representatives for the enforcement of obligations under the Directive”. Article 6 makes it possible to institute proceedings in the courts of the host Member State, even if this would not have been possible under the then applicable rules of article 5 (1) of the Brussels Convention. *50
In the Laval case, the protective ambit of article 5/6 is turned upside down by the boycott of Bygnadds. The posted Latvian workers have not been allowed to enter the building site and complete the contract. They were sent home and laid off from employment. A claim against their employer is theoretically possible but frustrated in fact by the industrial action of Bygnadds. Instead of the article protecting posted workers, exactly the opposite occurs: application of the article deprives them of their chance to work and eventually to claim Swedish wages (should they be applicable under article 3 (8) of directive 96/71) by using the mechanisms of article 5/6. This is in clear violation of the Wolff judgment of the ECJ *51 wherein the Court insisted that “Member States must ensure, in particular, that workers have available to them adequate procedures in order actually to obtain minimum rates of pay”.
The boycott is made possible by the Lex Britannia, which, as has been stated above, in itself violates EU law insofar as it allows discriminatory and selective action against service providers from EU countries and deprives posted workers of the rights to which they are entitled under directive 96/71. In reality, the boycott aims at closing the Swedish construction market to EU service providers and their posted workers. This is not justified under Community law.
As a subsidiary argument, Bygnadds may want to rely on article 3 (10) of directive 96/71, which allows the host Member State to apply to posted workers provisions of its laws, regulations, and administrative provisions governing matters additional to those set out in article 3 (1), provided that these additional rules fall within the category of ‘public policy provisions’. *52 Bygnadds and the Swedish government argue that the Medbestämmandelagen with the special rules of the Lex Britannia must be characterised as such public policy provisions because they enshrine the Swedish ‘social model’ in order to avoid social dumping.
Such an interpretation could possibly be based also on constitutional considerations as discussed above (see section 2.4). In Omega *53 , the Court linked the ‘public policy’ concept to protection of fundamental rights — in this case, the right to human dignity. The same should be true with respect to the relevance for public policy of the defence of social rights by collective action of trade unions, which are, as I have shown, guaranteed under article 28 of the Charter of Fundamental Rights and indirectly in article 136 (1) EC / article 11 ECHR.
But even under such an interpretation of the public policy provisions, it must be in conformity with the requirements of equal treatment and in compliance with the treaty. *54 This is not the case with the Lex Britannia, because of its discriminatory effect on EU-based enterprises in discarding the collective bargaining agreements concluded with the home trade unions. Second, public policy provisions must be construed strictly as an exception to the freedom to provide services under primary Community law, to which, as is shown by the legal basis of the directive, namely, the internal market provisions, it should contribute by improving competitive conditions, as opposed to creating new obstacles; therefore, national provisions invoking public policy rationale may not serve merely to protect economic goals (in this case, the national construction market’s action against alleged social dumping).
Third, in my opinion, paragraph 10 of article 3 of directive 96/71 only concerns provisions beyond the list of article 3 (1) — that is, other matters than minimum rates of pay. Insofar as the Lex Britannia enables Swedish labour unions to impose their national standards of wage bargaining on EU-based service providers, this is possible only under the criteria of article 3 (1) lit. c) respecting paragraph 8, which have already been discarded. The ‘public policy provision’ is not meant to give additional rights to MemberStates or their trade unions concerning pay rates beyond those of paragraphs 1 and 8 of article 3.
The Viking case is different from Laval insofar as directive 96/71 is not applicable here according to its Article 1 (2). Finland had not adopted specific legislation similar to the Swedish Lex Britannia as could be invoked to justify or condemn industrial action. Finnish law does not seem to contain any discriminatory element but protects the right to strike under non‑discriminatory conditions and limitations, including respect for directly applicable EC provisions (see section 4 above).
With regard to justification for the social action by the ITWF and FSU against Viking, some support can be found in the Überseering judgment. *55 Here, the ECJ allowed certain restrictions against companies moving into one jurisdiction, for social policy reasons. The case concerned a somewhat surprising consequence of the German seat theory. Here, a company formed according to the law of one state (the Netherlands) and whose ownership was transferred to another (Germany) had to be re‑established in the country of residence of its owners in order to have standing in civil litigation matters. It was a debated question under German law whether this consequence of the seat theory is really imposed by German law or not. *56 In a carefully worded decision, the ECJ repeated its insistence on the country-of-origin principle, whereby the company did not cease to be validly incorporated under Dutch law. On the other hand, it allowed certain limits under public policy criteria. Therefore, the Court aptly stated:
It is not inconceivable that overriding requirements relating to the general interest, such as the protection of the interests of creditors, minority shareholders, employees and even the taxation authorities may, in certain circumstances and subject to certain conditions, justify restrictions on freedom of establishment (paragraph 93).
The crucial question here lies in how far “overriding requirements […] such as the protection of employees” justified the social action taken by the ITWF and the Finnish Seamen’s Union, and whether their action was proportionate. In its reference, the English Court of Appeal wants to know whether the public policy proviso of article 46 EC can be applied in this case.
It should not be forgotten that the strike originated from the FOC policy of the ITWF, which includes a social objective in protecting seafarers against the practice of flags of convenience, which may lead — as the Viking case clearly shows — to an eventual ‘negative competition about wages’. The reference of EC law to several documents protecting social rights of workers included in article 136 (1) EC suggests that this is a legitimate objective that can — and eventually must — be enforced via social action, including strikes, in order to be effective. Whether this action violates the principle of proportionality should be answered by reference to the balancing test spelt out in Schmidberger. *57
In the latter judgment, the Court weighed the right to free movement on the Austrian Brenner Autobahn of the transportation company against the right of environmental organisations to peaceful demonstrations by blocking the Autobahn for a limited time. Fundamental rights as enshrined in articles 10 and 11 ECHR — e.g., freedom of expression and assembly — may justify restrictions of free movement, being “fundamental pillars of a democratic society” to which the EC and MemberStates adhere (see article 6 (2) EU). Fundamental rights may be limited for public policy reasons and must be viewed in relation to their ‘social purpose’, but these limitations in themselves must be proportionate. The ECJ applied a balancing test in the case before it: the demonstration was limited in time, scope, and intensity — it cannot be compared to thesituation in the caseCommission v. France *58 , where violence was used by wildcat demonstrators against Spanish fruit shipments that was even to some extent supported by the inaction of the French authorities. In Schmidberger, the demonstration was authorised by the regional government; the impediment to the business of the transportation company was not serious enough to give rise to state liability.
In the Viking case, the social action of the Finnish Seamen’s Union was in conformity with EU and Finnish law, as the Finnish labour court in the first litigation implicitly held by refusing an injunction against the FSU. This is in contrast to Laval, where the boycott was not in conformity with primary or secondary EC law and could not be justified by the Swedish Lex Britannia, which itself violated EC law. The action of the FSU and the boycott on the part of the ITWF was directed against only one part of the business of Viking — the reflagging of the Rosella — while the action against Laval made the provision of services by posted workers impossible. The Finnish action was covered by legitimate social policy goals as recognised in article 136 (1) EC and in conformity with the wide discretion allowed to Member States under the case law of the ECtHR in relation to article 11 ECHR. *59 The policy of the ITWF as mentioned in question 8 of the referring court
that vessels should be flagged in the registry of the country in which the beneficial ownership and control of the vessel is situated so that the trade union of the country of beneficial ownership of a vessel have the right to conclude collective bargaining agreements in respect of that vessel
therefore seems to be justified by social objectives and by the right to collective action as recognised in article 28 of the Charter. It does not carry any discriminatory element but applies to all vessels under a FOC. Strike action as such cannot be regarded as not being proportionate for attaining these legitimate objectives even if it may provoke losses of a company that is hit by such action — otherwise, labour unions would not have an efficient instrument to fight for their legitimate aims, such as social protection of workers. The ITWF and FSU therefore are allowed to use the threat, and eventual action, of a strike to implement their FOC policy. This must be true even if the strike prevents reflagging of the Rosella in the Viking case, which might eventually lead to substantial losses for the company, forcing it to abandon this part of the business. But this is the consequence of any social action and nothing particular to Viking. The proportionality criterion cannot be used to undermine the very substance of the right to strike. EC law is not qualified to solve this conflict by referring to a somewhat vague ‘proportionality’ proviso, unless the social action is directed against the ‘nationality’ of the ship-owner or serves merely to protect workers with Finnish citizenship, which, as the English Court of Appeal correctly held *60 , cannot be said to be the case here.
The opinion of Advocate General Mengozzi covers 42 pages, with 107 footnotes, and gives a very detailed overview of Swedish and Community legislation and social practice with regard to the conflict that arose in Laval. The AG takes as a starting point — just as this author does — the direct effect of article 49 EC on social action by labour unions (see paragraphs 155–161) and rejects the argument of the Swedish and Danish governments that the — admittedly fundamental — right to strike is “beyond the reach” of EC Law (paragraphs 233–236). A strike or boycott as in the Laval case must therefore be regarded as a “restriction on the freedom to provide services” (paragraph 233) that can, however, be justified by action to protect workers and to fight social dumping (paragraph 249), provided that it is in line with the proportionality criteria (discussed in paragraphs 245 et seq.).
The methodological starting point, however, of the AG differs from mine insofar as he interprets in great detail article 3 (1) and (8) of directive 96/71 and their de facto implementation via the Swedish system of collective bargaining. The Lex Britannia is therefore for him without problems, as it ensures the effectiveness of the Swedish system (paragraph 185), without being measured under article 49 and the mutual recognition principle. The core of the analysis is in paragraph 8, which enables Swedish trade unions not only to set minimum rates of pay as prescribed in paragraph 1 but to insist on full compliance with Swedish wage levels, perhaps under a somewhat lighter regime of tie-in conditions, without the target entity becoming a member of a trade association. This is, in his opinion, also justified under the minimum protection clause of article 3 (7). A conflict with directive 96/71 and later article 49 EC arises only where the wage rates as enforced by the Swedish trade unions also include certain contribution fees, supervision charges, insurance premiums, and the like that cannot be subsumed under the concept of pay or that are already covered by the home country jurisdiction (paragraph 281).
By imposing strict equality of treatment between foreign service providers and those domestic undertakings, the Swedish system appears in fact to disregard the very characteristics of the freedom to provide services by fully assimilating the temporary posting of workers by a service provider of a Member State to Sweden to a permanent activity carried on by undertakings that are established in Swedish territory.
The Swedish system, in imposing equal rates of pay in unequal economic conditions — namely, on the one hand, permanent work performed in Sweden on a regular basis and, on the other, temporary work done by posted workers under a service provider established in another EU country — amounts to indirect discrimination that should not be allowed under either article 49 or directive 96/71, which is a mere concretisation. The minimum protection clause should, as is the usual case in EC law, be interpreted strictly in the light of the fundamental freedoms of Laval to offer its services EC-wide and not serve as a basis for effectively sealing off the Swedish labour market.
The analysis of article 49 and especially the proportionality argument is more critical to the practice of the Swedish labour unions and shows more ‘sympathy’ with the situation of Laval.
In that connection, it seems to me that the fact of making the very possibility of applying a given rate of pay conditional upon prior signing up to all of the conditions of a collective agreement that apply in practice to undertakings established in Sweden in the same sector and in a similar situation goes beyond what is necessary to ensure the protection of workers and to prevent social dumping (paragraph 280).
As a result, the opinion gives a somewhat blurred picture: by generously interpreting article 3 (1), (7), and (8) of directive 96/71 in the sense of Swedish wage conditions, the competitive effect of the wage advantage of Laval by posting its workers who are under Latvian jurisdiction and bargaining agreements is neutralised. However, the way collective bargaining was imposed on Laval, on the one hand, and the excessive charges added to the wage package, on the other, as allowed by Swedish practice, seem to be in contradiction with the proportionality principle as enshrined in article 49 and therefore make the boycott illegal. One will have to wait to see how the Court finally decides.
The starting point of the opinion of AG Poiares Maduro is similar to that of AG Mengozzi as far as the relationship between the fundamental right to social action and the free movement rules, on the one hand, and the horizontal direct effect of article 49 respecting article 43 EC, on the other, is concerned. However, there are interesting nuances in their reasoning. AG Poiares Maduro argues mostly from a model of allocative efficiency (paragraph 33). Community law is aimed at “bringing these policies [e.g., social protection of workers via collective action rights of trade unions v. free movement rights in the area of establishment and provision of services] together”:
Therefore, the fact that a restriction on freedom of movement arises out of the exercise of a fundamental right or of conduct falling within the ambit of the social policy provisions does not render the provisions on freedom of movement inapplicable (paragraph 23).
The horizontal application of the fundamental freedoms of articles 43 and 49 is justified by a theory of mittelbare Drittwirkung (indirect effect on third parties of fundamental rights) as developed in German and other Member States’ constitutional law, and as implicit in the ECJ Commission v. France *61 and Schmidberger *62 cases. However, this argument is not quite correct, because the theory of mittelbare Drittwirkung is based on an obligation of the state (and its courts of law) to protect fundamental rights also in private law relations where no party autonomy exists, while the theory of horizontal effect as developed in this paper allows direct action of injured parties against collective regulations without any interference by the state; the judge hearing the case has no discretion but to enforce the free movement rights against violations by a sporting association or by trade unions. But the end results of both arguments are not dissimilar.
Since directive 96/71 is not applicable in the Viking case, and, since Finnish law does not contain legislation similar to the Swedish Lex Britannia, the case must be resolved according to a balancing test, which the AG set out in great clarity and brilliance (paragraphs 57–72). The question is how far trade unions can take social action against acts of relocation by undertakings that are protected by the free movement rules. On the one hand, workers (and their unions) must accept the
recurring negative consequences that are inherent to the common market’s creation of increasing prosperity, in exchange for which society must commit itself to the general improvement of their living and working conditions, and to the provision of economic support to those workers who, as a consequence of market forces, come into difficulties (paragraph 59).
This balancing, contrary to the reasoning set forth by AG Mengozzi and much of the case law of the ECJ that has been referred to by this author (in section 6 above), is performed not by applying a proportionality test but via the classical argument of market segregation:
A coordinated policy of collective action among unions normally constitutes a legitimate means to protect the wages and working conditions of seafarers. Yet, collective action that has the effect of partitioning the labour market and that impedes the hiring of seafarers from certain Member States in order to protect the jobs of seafarers in other Member States would strike at the heart of the principle of non‑discrimination on which the common market is founded (paragraph 62).
This seemingly simple test has a number of drawbacks, as the Viking case clearly shows. The collective actions of the ITF and FSU seemingly partition the labour market in attacking the reflagging and thereby preventing the hiring of (cheaper) Estonian seafarers. But, on the other hand, this is, in the eyes of labour unions acting in solidarity, justified by a general policy against flags of convenience, not against Estonian workers in particular. Who is to judge the legitimacy of this social policy even if it may have a detrimental effect on free movement? These effects may also be purely accidental and an unavoidable consequence of a social action if there is desire for it to be effective. The AG also seems to exaggerate the parallels between social actions in Commission v. France and Viking (paragraph 68): While the first made free movement impossible by wildcat actions blocking roads in France against Spanish fruit exporters, Viking can always run its Rosella (as it indeed has done for more than four years following protracted negotiations with the FSU, though perhaps without making the expected profit — not that making a profit is protected under the free movement rules!); these economic disadvantages of social action may also materialise in any other conflict concerning relocation of business within the EU. In my opinion, Viking cannot expect special protection against social actions established to impede reflagging, unless this action is in itself disproportionate (as in Commission v. France ), which, in my personal opinion, it is not.