When Efficiency Results in Redistribution: The Conflict over the Single Services Market

The discussion of the Services Directive from 2004 onwards showed an unprecedented extent of politicisation of a single-market issue. Coinciding with the Eastern enlargement in 2004, the easing of the services freedom through the Directive raised significant redistributive issues, given the differences in labour costs. The article analyses why mutual recognition is so controversial in services, by contrast with its acceptance in developing the internal market in goods. It is shown that the problem of trust in the regulatory systems of other member states that arises under mutual recognition is being addressed through administrative cooperation, but the problem of inequality between workers that arises from the application of home country regulations has not been resolved.

never roused much widespread political attention for goods, but for services the Commission was told differently.
By integrating segmented national markets, internal market legislation significantly enhances efficiency. Companies no longer need to adapt their goods or services to the different domestic regulations of member states. Instead, there are either common harmonised rules, or there is mutual recognition. This lifting of market segmentation allows companies to exploit economies of scale, while customers may enjoy greater product variety. Internal market policies are thus classic examples of measures increasing Pareto-efficiency (Majone 1989: 166-8;1992).
This article analyses why the realisation of the internal market for services was nevertheless perceived as a highly redistributive exercise. The explanation builds on the difference of services vis-a`-vis trade in goods, the specifics of governance through mutual recognition, and the increased heterogeneity among member states after enlargement. I argue that integrating markets via mutual recognition has implications for three relationships: for the relationship among EU governments, which implicitly delegate regulatory authority between themselves; for the relationship of governments to their citizens given that they are politically responsible for market regulation; and finally among different EU citizens who are subject to the different regulations of their home countries while engaging in the same activities.
For mutual recognition to work in services, member states have to perceive themselves as cooperating rather than competing entities. Integration becomes acceptable only if rules are being set and controlled according to domestic criteria, and not simply with a view to outcompeting other member states. If integrating markets via mutual recognition invites forumshopping, redistributive issues are in the forefront. The member state which manages to make its rules the most market-friendly gains the most. For redistributive issues, the Union has insufficient input-legitimation (Scharpf 2004). Even if liberalising services markets strengthens general welfare, if liberalisation also has highly redistributive effects this is unlikely to be perceived as legitimate, due to the lack of a common demos showing sufficient solidarity.
The article starts by analysing the original proposal for a directive. Then it approaches the issue of why mutual recognition proved so contentious by taking a closer look at the regulation of services trade in the EU. A discussion of the German experience with mutual recognition in services exemplifies the resulting redistribution, leading to an analysis of the limits of mutual recognition in services. Finally, it shows how member states have tried to contain redistributive issues in services trade, focusing on the unilateral and bilateral German responses as well as the compromise on the Services Directive. Paradoxically, in services the choice of mutual recognition over the harmonisation of rules seems to imply that member states will need to follow common administrative procedures in implementation to lessen redistributive consequences. However, the inequality that mutual recognition in services imposes on citizens cannot be fully contained in this way and makes its application contentious.

The Bolkestein Directive
Launched in early 2004, the draft Directive aimed at realising the internal market for services in all those areas where specific legislative measures had not yet been taken (as in the case of financial services, for instance). Due to the importance of services, the Commission targeted about 50 per cent of all economic activities of member states with this single directive. 1 The Directive aimed at realising both the freedom of establishment and the freedom of services, exempting only lotteries and all genuinely public services with no profit interest (e.g. education, cultural activities). Health and social services were included. In order to achieve its ambitious goal, the draft Directive relied on the principle of home-country control. Member states were required to mutually recognise services regulated in other member states as equivalent to domestically regulated services and to abolish excessive regulatory requirements. 2 In order to support the necessary cooperation between home-and host-country authorities, the Directive obliged national authorities to cooperate with each other. Thus, the possibilities for host-country authorities to obtain information from home-country authorities as to the legality of companies posting workers would be greatly improved.
Given the highly regulated nature of most services, the deregulatory potential of the Directive was considerable, as was aptly described by the former Commissioner Bolkestein: We cannot expect European businesses to set the global competitiveness standard or to give their customers the quality and choice they deserve while they still have their hands tied behind their backs by national red tape, eleven years after the 1993 deadline for creating a real Internal Market. Some of the national restrictions are archaic, overly burdensome and break EU law. Those have simply got to go. A much longer list of differing national rules needs sweeping regulatory reform. 3 Since sector-specific attempts at building the internal market for services had already proven cumbersome, the Directive was bound to be controversial. With its broad scope, it was hardly possible to assess all the implications of the Directive. Moreover, the Services Directive explicitly complemented existing services law, with uncertain implications resulting from the interaction. This overlap concerned in particular the Posted Workers Directive (96/71/EC) of 1996. The services draft foresaw the easing of some restrictions on posted workers, like the need to carry papers for local controls in the host country and the obligation to appoint a national representative, making control by the host country more difficult.

The Conflict over the Single Services Market 849
While the Commission tried to advertise the services proposal as a measure simply enhancing efficiency in a Pareto-optimising sense, the reactions to the draft emphasised its redistributive consequences. Why the liberalisation of services trade is more closely linked to redistribution than to Pareto-efficiency will be discussed below.
The Puzzle: Why do Services Pose Different Problems than Goods?
Why, then, was the transfer of a principle well established in goods trade to services so controversial? To approach this question, it is necessary to discuss both the nature of mutual recognition and of services regulation and trade. I will then turn to the German experience with mutual recognition in services to illustrate the implications for redistribution. The section closes with a discussion of the limits of mutual recognition in services.

Mutual Recognition and the Regulation of Services Trade in the EU
While mutual recognition was already considered by the Commission in the 1960s as a way to integrate tax policy (Genschel 2007), it entered the scene noticeably only in 1979, with the Cassis ruling (C-120/78) of the European Court of Justice (ECJ). Drawing on Dassonville (C-8/74), the ECJ gave a broad meaning to the freedom of trade in goods (Article 28), finding that goods legally marketed in one member state can also be marketed in all other member states. This is the obligation to mutually recognise goods from other member states, if they conform to the rules of their home country. However, by broadening the reach of the market freedoms under the Cassis de Dijon case law, the ECJ simultaneously enhanced the possibilities of member states to claim exceptions beyond those already foreseen in the Treaty by Article 30. Member states in principle retain the right to regulate their domestic markets in sensitive areas, by invoking mandatory requirements goods have to adhere to. As regulation needs to be proportionate, and the Court takes a very strict view of proportionality, the scope is however quite circumscribed (Hatzopoulos and Do 2006: 965f).
As is well known, the Commission readily took up the idea of mutual recognition and launched the internal market initiative around it. While there was some discussion in the legal profession that goods regulated according to foreign rules would have to be mutually recognised (Alter and Meunier-Aitsahalia 1994), the issue did not become politicised as it was subsequently to do in services.
Integrating markets with mutual recognition evades the difficulty of agreeing on harmonisation. However, compared to harmonised rules, mutual recognition is much more difficult to implement. Companies may believe that their goods and services are regulated in an equivalent way and therefore qualify for mutual recognition -it may also happen, however, that national authorities are of a different opinion (Pelkmans 2007). As the authorities responsible for the control of market regulations have to decide whether the rules of the 26 other member states are equivalent or not, mutual recognition entails significant transaction costs. The solution of conflicts is thus shifted from the decision-making to the implementation stage (He´ritier 1999: 16-19, 23). Normatively, mutual recognition also has specific advantages. Joerges makes the important argument that by having to honour the regulations of the other member states, mutual recognition manages to lessen the legitimatory deficits of national policy-making. These deficits arise as national legislation focuses only on the domestic situation despite having significant externalities for other member states in an integrated market (Joerges 2006: 790).
Following a similar logic to the one of the internal market, the European Council in Tampere decided in the late 1990s to apply mutual recognition to the area of justice and home affairs (JHA). Again, there was a perceived need for cooperation, but an inability to agree on harmonisation (Lavenex 2007). Thus, even for rules which normally fall under parliamentary prerogative, mutual recognition may be acceptable. If mutual recognition in services appears to be more difficult than for goods it is therefore unlikely to be simply due to the fact that services regulation is politically too sensitive to be mutually recognised.
To sum up, mutual recognition can act as an alternative to harmonisation if member states differ in their -equivalent -regulations. Then they could accept each other's rules instead of bothering to agree on common ones. Key to the acceptability of the principle is the question of what is regarded as 'equivalent'. If rules have to attain the same qualitative degree of regulation, the deregulatory push following from mutual recognition is much less than when merely 'adequate' regulation has to be recognised, mirroring only essential requirements. Even though the ECJ largely follows this latter view, it was only with services that the issue became broadly politicised. To further the understanding of why the question of equivalence versus adequacy was so much more contentious in services, it is necessary to proceed by looking at the specifics of services trade, regulation, and the legal provisions for the freedom of services.
In contrast to goods, only a few services can cross borders independent of their production. It is only for these correspondence services (for instance financial services, telecommunications) that services trade largely resembles goods trade. For most services, the delivery coincides with consumption, meaning that either the service provider (exercising active freedom of services) or the service consumer (exercising passive freedom of services) has to move for services trade (Hailbronner and Nachbaur 1992: 108). It is on the problems of the active freedom of services that this article focuses.
Services are in a certain sense invisible, which is why it is often difficult to separate their production from their consumption. This makes regulation much more constitutive for services than for goods. Regulation can concern market access (e.g. certain training requirements), operation (e.g. solvency The Conflict over the Single Services Market 851 requirements, speed limits), the products themselves, and their distribution (cf. Roth 2002: 16). In contrast to goods, the regulation of services relies heavily on what one can compare to process standards (cf. Troberg 1997. It follows that 'there is a closer connection between services regulation and labour market regulation than in the case of goods' (Pelkmans and van Kessel 2007: 7).
We know from goods trade that product and process standards are subject to different kinds of competitive pressures (Scharpf 1999). While consumer demand for high-quality products may keep product standards up, process standards are much more subject to competitive pressures. Since process standards are more relevant for services than for goods, it is much more likely that producers will make a strategic choice of home country simply for ease of regulation, i.e. that they will engage in so-called forum shopping. This is facilitated by the fact that compared to goods the investment and commitment to a location is often lower in services.
To what extent does EC law already apply the home-country rule to services provision? It is necessary to ask this question to get an idea of whether the Services Directive merely codified the case law of the Court or expanded on it. The services freedom is concerned with remunerated, temporary services delivery across borders (Roth 1988: 41). It can therefore be distinguished from the freedoms of labour and of establishment. If someone occasionally works in another country, they will probably profit from the freedom to provide services; if they do it on a continuous basis with some sort of establishment, it is the freedom of establishment that matters. If an EU worker is engaged in another labour market, it will be the free movement of labour. These freedoms are differently regulated: with the freedom of establishment and of labour, companies and persons are regulated on a par with nationals in the country where the services are provided. But if the services freedom is evoked, the regulations of the country of establishment (i.e. the home country) and not so much of service provision (i.e. the host country) are applicable. However, host countries can apply rules that are covered by the general interest -provided such rules have not already been observed in the home country (Hailbronner and Nachbaur 1992: 112).
In contrast to the freedom of goods, for a long time the ECJ interpreted the freedom of services in quite a restrictive way, not covering continuous, regular activities (Hatzopoulos 2000: 63f;Roth 2002: 20;Davies 2007b: 14). Article 50 clearly prescribes the host-country rule for services trade and restricts it to temporary activities. This is exemplified by the case law on the posting of workers, where the ECJ allowed France in the Rush Portuguesa case (C-113/89) to apply its minimum wages to workers temporarily posted from Portugal. This ruling led to the Posted Workers Directive. More recently, however, the ECJ has pursued a more liberal approach with regard to services, emphasising the need to eliminate hindrances to services trade more than the right of the host country to impose its regulations. At the same time, the number of Court cases concerning the freedom to provide services is increasing, showing its growing relevance (Hatzopoulos and Do 2006: 923). Thus, in 2003 the ECJ did not stress the temporary nature of services in the 12 May 2005 Schnitzer case (C-215/01), loosening the relationship with the freedom of establishment. In a later case, the ECJ went so far as to state that 'all services that are not offered on a stable and continuous basis from an established professional base in the Member State of destination constitute provision of services within the meaning of Article 49 EC' (Hatzopoulos and Do 2006: 929, emphasis original). This interpretation enhances the scope for mutual recognition by emphasising the right of companies to forum-shop, thereby increasing competitive pressure arising from differences in services regulation. The proposal for a Services Directive took up this incipient change in the case law, seeking to codify it in a radical way (De Witte 2007: 9f). However, the Treaty's text and its interpretation meant that member states had a good legal basis to object to the radicalised home-country principle of the Directive.

Redistributive Concerns: The Contention about the Draft Directive Illustrated with the German Case
Given that services were part of the internal market initiative of the late 1980s, the Services Directive was long overdue. With services' growing relevance, the failure to establish the internal market implied significant efficiency losses. 4 Why did redistributive worries then overshadow efficiency concerns? The particular timing of the draft Directive in early 2004 was important. Within a few months of the proposal, Eastern enlargement significantly increased the heterogeneity of the EU. This fundamentally altered the basis of a regime built on home-country rule. Governance based on mutual recognition requires equivalent rules and administrative cooperation. While in the EU-15 mutual recognition was already ridden with prerequisites (Pelkmans 2007), in the EU-25 (27 to be) it could only become harder. Redistributive effects arise on the one hand due to the labourintensive nature of most services. Eastern enlargement implied that the lower wages of these countries immediately exert pressures, most of all in those countries relying on collective wage agreements instead of minimum wages, as these are not automatically binding for service providers from other member states.
The German case illustrates well how redistributive issues overshadowed the promise of efficiency gains. Germany had joined most other member states (with the exception of the UK, Ireland and Sweden) in using the transition provisions (lasting up to seven years) to restrict the freedom of labour for the East European new member states. In addition, Germany negotiated a transition regime for the freedom of services for some sectors. The German public was therefore surprised when, a few months after enlargement, East Europeans nevertheless put significant pressure on the The Conflict over the Single Services Market 853 national job market -simply by using the services freedom. It is important that Germany does not have a generalised minimum wage it can impose (Christen 2004;Temming 2005). 5 Under the services freedom, workers can come in temporarily -which is interpreted as up to one year -and replace German workers, while working for the wages of their home country.
Most noted in the press was the case of the slaughterhouses, where Germans were laid off in large numbers as East European service providers were brought in, working for little money under deplorable working conditions. As a result of this experience, it was feared that the Services Directive would bring similar pressure to other sectors. 6 The situation was complicated as East Europeans could work in Germany under different legal provisions, either relying on the freedom of establishment or of services, added to which were illegal activities. 7 Under the freedom of establishment, East Europeans faced no restrictions but they had to comply with German laws. No specific wage and social security obligations exist for establishments, inviting social security fraud, for instance through mock self-employment.
Under the freedom of services, East Europeans can be posted from an Eastern European company to deliver temporary services in Germany, except in the exempted sectors of construction, cleaning, and internal decoration (Temming 2005: 188). Following the Posted Workers Directive, German labour conditions apply for all branches, but, since there is no general minimum wage, there are no restrictions on what posted workers have to be paid. The posting company has to discharge social security expenses in the home country, where it also has to be active -mere mailbox companies are illegal, as are posted workers that are fully integrated in the German company's work process. A host of different possibilities for illegal activities result from these requirements. But violations are difficult to detect by the host country, whose authorities have to trust the controls of the home country. Moreover, there are tricky legal questions: it is probably insufficient if a company employs one person for recruitment and control purposes in, say, Poland, and posts 99 workers into Germany. But how many persons have to be employed in Poland? What proportion of the annual turnover has to be achieved in the home country in order to be seen as a company active there (Fleischwirtschaft 2005: 10)? A Commission 'practical guide' for the posting of workers suggests that companies should achieve a minimum of 25 per cent of total turnover in the posting state, where they should have been established for at least four months, with other cases requiring 'individual attention'. 8 With high unemployment rates in many new member states, old member states distrust whether new member states abide by the rules.
The German situation was particularly difficult due to not having a minimum wage, which is also the case in a few other member states (Denmark, Finland, Italy and Sweden). Nevertheless, the lesson told by the German case is a more general one. It shows the problems of applying home-country control to services trade, of which the ample opportunities for illegal activities are one important part. With the significant wage differentials in the EU after enlargement, a greater emphasis on homecountry regulations for services has a significant deregulatory potentialwhich, after all, was partly wanted, as is evident in the citation from Commissioner Bolkestein, noted above. The contentious ECJ decisions in Laval (C-341/05) and Viking (C-438/05), at the end of 2007, have emphasised the competitive pressure (Joerges and Ro¨dl 2008), as have the Ru¨ffert (C-346/06) and Luxembourg (C-319/06) cases in 2008.

The Limits of Mutual Recognition in Services
Why, then, is mutual recognition for goods regarded as efficiency enhancing while for services it is perceived as a redistributive issue? After all, the rationale for trading both goods and services lies in exploiting comparative advantages. This section analyses why mutual recognition is more difficult to accept for services than for goods. Most services differ from goods, we saw, in that provision and consumption coincides, making it necessary that the service provider travels along. If the service provider has to become active in the host country, what does this imply for mutual recognition and home-country control?
Mutual recognition relies on the assumption that member states regulate markets differently, but in functionally equivalent ways. It implies that member state governments accept regulations on their territory for their population that were decided and legitimated in other member states (Schmidt 2007). Mutual recognition thus fundamentally concerns three interdependent relationships: the relationship among member state governments, the relationship of governments to their population, and the relationship between EU citizens who are subject to different regulations. By enquiring into these three relationships, we can see why mutual recognition is problematic in services.
Governments accept other governments' regulations only if they are equivalent, because they remain politically responsible to their population for the regulation of their markets. If rules are not equivalent, host-country rules apply, fragmenting the market, so that harmonisation is needed. Given their political responsibility, governments therefore have to trust each other as to their equivalence of setting and controlling regulations when agreeing to mutual recognition. It is here that the difference in the regulation of services compared to goods becomes relevant. For goods, in general, governments only have to trust each other to maintain sufficient regulation and control of product standards. This is supported by the interest of governments in the well-being of their own populations, and the interest of manufacturers in their reputation, which extends to exported goods. In the case of services, however, process standards have to be controlled which only in part affect the quality of services. Where services are being exported using the competitive advantage of lower wages, this requires a higher The Conflict over the Single Services Market 855 degree of trust. Governments have to trust that their counterparts behave altruistically and control service providers simply for the sake of other member states (cf. Scharpf 1997: ch. 4). This contrasts with governments' interest in effective controls for goods. The amount of fraud occurring with services trade shows how demanding controls are, and that governments have good reason to be sceptical about mutual recognition in services.
Secondly, for the relationship of the government to its own population, it is relevant that the increased legal certainty achieved for service providers under home-country control comes at the cost of significant legal uncertainty for consumers in the host countries. Mutual recognition implies that they no longer know under which rules services are provided temporarily in their country. This produced fears of competitive pressures on domestic regulations, as well as insecurity for final consumers who do not know which rules service providers would have to abide by and which rights they would have themselves as consumers (Nicolaı¨dis and Schmidt 2007).
More important yet, governments can no longer guarantee equal treatment to their citizens, as becomes apparent when turning to the (third) relationship between EU citizens. As service providers travel along with the traded services, it follows that very differently regulated service providers might work side-by-side simultaneously, raising important issues of equality. Davies puts the problems succinctly: The presumption is that service providers are exempt, above the law of the territory where they operate, and the rebuttal of that presumption is hard. The situation where competing service providers on a territory are subject to different legal regimes -that they essentially bring their own legal regime with them -becomes the usual one, with all the associated challenges to equality and competition norms. (Davies 2007b: 8) Davies therefore argues that home-country control for services violates the prohibition of nationality discrimination of Article 12 of the Treaty, as nationals of the host state are being discriminated against. Host-country control, in contrast, would generally lead to less inequality (Davies 2007b: 8).
An individual who is present in the jurisdiction but not subject to its regulation, and operating under a more beneficial regime, is a direct challenge to the content of citizenship -national or European -and its associated guarantees of equality and privilege. His domestic competitor sees his most privileged position as a national citizen undermined, while the two competitors, working side by side, operate under different legal regimes with different rights, despite a shared EU citizenship. (Davies 2007b: 7) To sum up, services are particularly rule-dependent, and normally cannot be traded without their providers. If mutual recognition is applied, this can give producers much greater scope for forum shopping. Mutual recognition in services is problematic with regard to three relationships. Among governments, it requires trust in an altruistic orientation, as service providers are controlled by a different jurisdiction than the one where they operate, facilitating illegal activities. In the relationship of governments to their population, governments might have to discriminate against their own citizens. Among EU citizens, mutual recognition raises issues of equality, as in the same situation EU citizens are subject to different rules even though they may be working side by side in the same workplace.
Mutual recognition for services, we can conclude, cannot be treated in a parallel way to goods. To accept mutual recognition in services, member states need to be assured among themselves that the delegation of competence inherent in mutual recognition is taken up responsibly. Since mutual recognition implies that differently regulated EU citizens may work side by side, the duty to take into account the interests of other member states, moreover, demands from member states to discriminate against their own citizens. To balance both (domestic and EU) interests by drawing appropriate borders as to the necessary extent of mutual recognition is inherently difficult for services.

Containing Redistributive Conflicts in Services Trade
Member states reacted to the redistributive effects of applying mutual recognition to service trade both by negotiating significant changes to the Services Directive and by adjusting their own laws and administrative practices. This section looks first at the autonomous response in Germany, which was particularly vulnerable to the effects of the services freedom, and then turns to the negotiations over the Directive.

German Reactions
One notable result of the debate over the Services Directive was to heighten the discussion about minimum wages in Germany, which started with the Red-Green coalition's agreement in May 2005 on extending the German posted workers law to all sectors of the economy. 9 This discussion has been continuing ever since. Since the state traditionally leaves wage agreements to the unions and employers' associations, a statutory minimum wage implies a significant institutional rupture. But it would be the easiest way to handle some of the redistributive issues arising from the services freedom, as well as responding to the general erosion of the German wage system. The discussion on minimum wages, which is ongoing, is therefore an interesting example of a Europeanisation effect (Schmidt et al. 2008).
Moreover, Germany has attempted to fight illegal activities and engaged in bilateral talks with its neighbours about the interpretation of the freedom of services. 10 Both reactions have been part of the mandate of the 'Task The Conflict over the Single Services Market 857 Force zur Beka¨mpfung des Missbrauchs der Dienstleistungs-und Niederlassungsfreiheit', 11 which was created in March 2005 to combat the adverse effects of the freedom of services and establishment after enlargement. Germany requires a postal address of the posting company as well as a translation of relevant documents concerning working contract, working times and pay into German (TAZ 2007). The measures envisaged in the draft Services Directive to facilitate the posting of workers clearly ran counter to these control efforts. Pursuant to its aim of easing the posting of workers, the Commission started an infringement procedure in late 2004 against Germany for restricting the services freedom disproportionately. But in its judgment of July 2007, the ECJ assessed the German interest in workable controls by requiring translated documents as not interfering with the services freedom (Commission v Germany, C-490/04). It remains to be seen whether this will lead the European Commission to back down from its criticism on how Germany, and other member states, have implemented the Posted Workers Directive, and to refrain from further liberalisation. As late as June 2007 the Commission published a communication on the Directive, criticising member states for overly controlling the exercise of the services freedom. 12 Moreover, since April 2006 Germany has compelled other member states to send a copy of all E101-forms (documenting social-security contributions in the home country) of workers posted to Germany to the Deutsche Rentenversicherung in Wu¨rzburg. 13 Previously, Germany suffered a serious setback in its activities to combat fraud when the ECJ ruled in early 2006 that member states have to accept existing E101-forms, even when they are obviously improperly issued (Herbosch Kiere, C-2/05). In cases of suspicion, member states have to contact the issuing authority of the concerned member state or start an infringement procedure (Article 227 EC Treaty) against each other. But national courts may not unilaterally reject administrative acts of other member states. This left Germany relatively helpless in a case where Germans had employed Portuguese workers on a permanent basis, pretending that they were being posted from Portugal, thus evading social security payments in Germany. 14 Aiming at a 'common understanding' of the freedom of services, Germany has conducted bilateral talks with other member states, both old (e.g. Denmark, the Netherlands, Austria) and new (Poland and Hungary) on a regular basis (BMF/BMAS 2006: 4-6). At issue are the criteria for according the status 'posted worker' and the conditions for giving out E101-certificates, including the required economic activity in the country of origin (combating mailbox companies), and the distinction between real and mock self-employment. 15 Moreover, the bilateral talks have allowed the investigation of specific problems, such as the situation in German slaughterhouses. 16 The talks explicitly aim to strengthen mutual trust (BMF/BMAS 2006: 4) and seek the signing of administrative cooperation agreements between Germany and the new member states (which exist already between Germany and France, between France and Belgium, and between Britain and the Netherlands) in order to improve cross-border controls and cooperation between the respective authorities. 17 The Fate of the Bolkestein Directive Redistributive rather than efficiency issues also dominated the negotiations over the Services Directive. The compromise which was reached in the European Parliament between the Social Democrats and Christian Democrats abolished all references to the contentious home-country principle, speaking only of the obligation to enable the freedom of services. 18 However, a list of measures is included which member states may not impose, such as special duties to register in the host country or ex ante certification as well as prescriptions as to materials and tools used. So in all these cases home-country rules apply, although the Directive refrained from saying so openly (Nicolaı¨dis and Schmidt 2007). Importantly, the list of justifications for host-country requirements in Article 16 III is much narrower than the ECJ case law, implying a 'deregulatory shift' (De Witte 2007: 12;Davies 2007b: 12, 18).
The final Directive is more restricted in scope than the draft, exempting health services, utilities, public transport, social and security services, temporary workers, gambling and lotteries, waste, audiovisual services, electronic communication, and financial and legal services. With regard to consumer protection, the Directive applies host-country rules (Article 3 II). Also, the easing of controls on the posting of workers was removed, much to the displeasure of the East European member states. Here, the Commission promised a separate follow-up on the workings of the Posted Workers Directive, as mentioned above. It remains to be seen whether the recent ECJ judgment concerning the German controls brings the Commission's attempts to liberalise the posted workers regime to a halt.
Other provisions remained more or less untouched. Thus, member states have to establish points of single contact for service providers in their administration, to abolish disproportionate regulatory burdens, and to allow service providers to do all formalities electronically. Importantly, the Directive includes far-reaching provisions on administrative cooperation (Chapter VI; Articles 28-36), detailing the responsibilities of the administrations in the home and the host country. In contrast to the original version, host-country authorities will now be responsible for controlling those rules which they impose themselves (Article 31). But, generally, homecountry authorities are the ones legally responsible for oversight. Both authorities must cooperate closely: since home-country authorities cannot operate in the host country, they will have to request the host authorities to act. Notably, the Directive establishes a duty to cooperate among the member states' administrations. This has not existed before to such an extent. 19 Thus, Article 28 (8) foresees that the Commission will start an infringement procedure if member states fail to comply with their duties of The Conflict over the Single Services Market 859 'mutual assistance'. In order to facilitate cooperation across language barriers, the Commission is promoting an information system providing for automatic translation of specific standardised paragraphs.
Thus, the Services Directive lays the foundation for administrations to operate transnationally. Instead of simply following the instructions given in their national hierarchy of command, with the Minister on top being ultimately politically responsible (Do¨hler 2001), administrations now also have to comply with horizontal demands, originating in other member states' administrations.
In the negotiations on the Directive, the administrative changes required received relatively little attention. They impose, however, major institutional changes. '[T]he real importance of the Directive is elsewhere. Its proper title should perhaps be ''the Directive on harmonisation and modernisation of public administration''' (Davies 2007a: 239). The requirement to provide for points of single contact implies a paradigmatic change for administrations, as these now have to be thought of from the point of view of the citizen accessing the administration rather than from the point of view of state organisation and lines of internal administrative accountability and responsibility (Schliesky 2005: 891). The obligation of administrative assistance among member states intensifies these changes, and raises questions as to which administration acts under what kind of law. Thus, if the host-country administration acts on behalf of the home-country administration, it would have to do so on the legal basis of the home country. But how would legal protection against these acts be organised, which court would subject the administration under what kind of law? Could a home-country court, say in Estonia, order an administration, say in the UK, to postpone or alter its acts? How would different data protection laws be handled? It is likely that the duty to cooperate will lead to further harmonisation of administrative laws of member states, as national laws have to be adapted to the new requirements (Schliesky 2008: 223-8). Thus, the Directive presents a significant break from the past principle that member states are free to implement European directives with the European Union not interfering with member states' administrative organisation (Schliesky 2005: 893f).
To sum up, redistributive concerns were appeased by reducing the scope of the Directive and by abolishing the home-country principle, at least officially. By the back door, however, mutual recognition and home-country control remain. The Directive does not provide an answer to the resulting issues of (in)equality of service providers working side-by-side under different regulations and the obligation of member states to possibly discriminate against their own population. For the problem of homecountry control of service providers active in another member state, the Directive introduces an obligation of cooperation. The centrality of administrative cooperation is also apparent in the bilateral initiatives of the German government. In the case of services, regulation matters so much for competitive performance that incentives are high to engage in regulatory arbitrage or even fraud. The harmonisation of administrative procedures to contain these redistributive issues logically follows.

Conclusions
Mutual recognition is an alternative to integrating markets via harmonisation. For services markets it poses particular challenges. While integrated markets can further Pareto-efficiency, the specific rule-dependence of services, where production generally cannot be separated from consumption, implies significant redistributive consequences. Often little investment and long-term commitment is needed for services trade, making forum shopping quite easy. I have argued that to understand why mutual recognition in services has proved so contentious, three interdependent relationships have to be taken into account: among member states, between governments and their citizens, and among differently regulated EU citizens.
With mutual recognition, governments effectively delegate the regulation of services among each other. For services, member states have to trust each other to control sufficiently -even when just for the sake of customers abroad. In contrast to goods, in the control of services it is often possible to distinguish whether the beneficiaries of controls are domestic customers or those of another member state. Particularly in view of high unemployment figures in new member states, old member states are sceptical about whether the administrations of new member states will act altruistically to protect workers in old member states at the expense of their own population. Because of the difficulty of controlling service providers under homecountry rule, mutual recognition in services favours illegal activities.
In order to ameliorate such concerns of not playing by the rules, and thereby raising redistributive issues, the Services Directive strengthens administrative cooperation. As we saw, this is also what Germany attempted in bilateral talks. Interestingly, mutual recognition thus leads to an incipient harmonisation of state organisation. Traditionally, it is up to member states how they implement the law of the European Union. Mutual recognition allows 'unity in diversity' when it comes to regulation (Joerges 2006: 790), but, paradoxically, it leads to a new form of harmonisation in implementation in the case of services. In terms of institution-building, this is a significant development. Possibly, it lays the foundation for a network of nationally-based administrations pursuing European aims, rather than segmented national authorities pursuing domestic goals (cf. Egeberg 2008). It remains to be seen to what extent the redistributive conflicts inherent in services trade can be mediated successfully on this level.
While administrative cooperation may strengthen trust among governments, the problems that mutual recognition in services raises in the relationship of governments to their citizens and among EU citizens are much more difficult to solve. As home-country rules are being taken along The Conflict over the Single Services Market 861 with services provision, these rules take effect outside the territory where they were legitimately enacted. Thus, service providers simultaneously work side-by-side under different rules and member states may be forced to discriminate against their own citizens with their domestic regulation. This is a consequence which is difficult to accept, and the Directive does not provide any answer to this problem of equality.
As Joerges (2006) argues, mutual recognition has the normative benefit of forcing member states to take into account their respective interests, given that economic decisions of one member state very likely have externalities in the single market. Mutual recognition works under the assumption that regulations are equivalent. Often, there are different ways of achieving the same regulatory objectives. By agreeing to mutual recognition, member states implicitly accept that the efficiency gains of a larger market come at the cost of some loss of capacity to pursue their own citizens' interests. This is the case whenever mutual recognition is applied. The example of services trade shows that it is not so much the kind of rules that are recognised, but the occurrence of very open inequality between different EU citizens that makes mutual recognition politically so contentious. With goods, and also with mutual recognition in JHA, EU citizens subject to different homecountry rules do not find themselves in the same place at the same time, noticing directly their different treatment. For services this is the case. It follows that, while member states have to take account of the regulatory situation in other member states, it is politically difficult for the Union to demand that member states accept rules that openly disadvantage their own citizens. Consequently, in those areas where mutual recognition leads to a direct comparison of EU citizens being subjected to different home-country rules, mutual recognition is bound to be contentious.