Liberal Intergovernmentalism,spillover and supranational immigration policy

Abstract

That the Lisbon Treaty lays the foundation for a supranational asylum and immigration policy is surprising, even more so for Liberal Intergovernmentalism (LI), whose founder Andrew Moravcsik predicts that no such development will take place. While the article uses LI as its point of departure, it shows that it runs into problems with regards to the policy area of asylum and immigration. The article therefore turns to the (neo-)functionalist concept of spillover. While working with the concept, it was deemed necessary to create a more coherent typology of different spillovers. The article suggests that the concept of spillover may be both descriptive and explanatory. With regards to descriptive spillover, it seems valuable to differentiate between widening and deepening spillovers, but concerning explanatory spillovers, more options became visible: there are unintended or intended functional spillovers, as well as unintended political, cultivated and social spillovers. The argument is illustrated through a detailed study of Sweden – a ‘reluctant European’ that within the area of asylum and immigration made a fundamental U-turn with regards to a supranationalism, a change that can be described as a social spillover.

Keywords

Asylum policy European integration immigration policy Liberal Intergovernmentalism spillover supranationalism

Introduction

Issues of asylum and immigration have long been considered as touching upon what a state is (Brochmann, 1996: 5; Collinson, 1993: 3; Donnan and Wilson, 1999: 15). Hannah Arendt even claimed that: ‘Theoretically, sovereignty is nowhere more absolute than in matters of emigration, naturalization, nationality and expulsion’ (Arendt, 1973: 278). It is then puzzling when states move from a limited intergovernmental co-operation to creating a constitutional foundation for supranational co-operation encompassing almost all asylum and immigration policies. Such a puzzle becomes even more perplexing when it concerns states, like Sweden, that traditionally have been viewed as a generally ’reluctant European’ (Johansson, 2002: 295; Miljan, 1977; Tallberg, 2001: 28f.).

Such a general Swedish reluctance to supranationalism has long been evident with regards to asylum policies. While a 1936 Official Report from the Swedish Government stated that both states and refugees will gain from international co-operation on asylum, and that a state will be more likely to make concessions if other states do the same, it nonetheless concluded that ‘Nevertheless … from the Swedish side one ought to not take on obligations, which aim at restricting the liberty to implement the immigration policy, deemed appropriate at each particular point of time’ (SOU, 1936: 53–57). This position was maintained for more than 60 years, but changed a few years into Swedish European Union (EU) membership when Sweden became an advocate of decisions with qualified majority voting (QMV). This article explains why.

The eminent scholar Andrew Moravcsik suggests that, since the mid-1990s, the EU has reached a ‘stable constitutional equilibrium’, an ‘institutional plateau’ (2005, 2008). One argument is that despite intense activity in areas such as defence, law and order, social policy, asylum and immigration, ‘…EU policy plays a subordinate role. EU policy in these areas tends to proceed by unanimity, with a subordinate role, if any, for the Commission, Parliament and Court’ (2005: 365, 2008: 174). In areas such as criminal law and policing, the actual state of affairs is more complicated than this (see Peers, 2011: 42–43), but in the area of asylum and immigration, the description is simply wrong.

Firstly, concerning the way decisions are made, the current Treaty on the Functioning of the European Union (TFEU) stipulates that QMV in the Council and co-decision by the Parliament is to be applied to the issues of asylum and immigration (Art 78 and 79). Secondly, with regards to what the EU can actually decide upon, the member states and the EU within this area have ‘shared competence’ (Art 4(2)(j)). This implies that, ‘the Member States shall exercise their competence to the extent that the Union has not exercised its competence’ (Art 2(2)). Thus, if the EU exercises its competence diligently, it may in principle result in complete harmonisation. What hinders this is that there are often more precise rules within each policy area. However, concerning asylum, immigration and external border controls, the EU has the power to ‘frame a common policy’ (Art 67(2)). The EU then has the right, through supranational means such as Council decisions with QMV and the Parliament co-deciding, to remove Member States’ competence in these areas. Concerning immigration, there is one interesting explicit exception stating that the Member States shall maintain the right to ‘determine volumes of admission of third-country nationals coming from third countries to their territory in order to seek work, whether employed or self-employed’ (Art 79(5)). However, this exception concerns a minor part of all immigration, of the circa 2.3 million new residence permits issued to non-EU citizens in 2009, only 27.5 per cent were granted for employment, with the remainder for family reasons, educational purposes, international protection and so on (Eurostat, 2011).

As such, although it largely remains for the EU to more comprehensively exercise its powers in the area of asylum and immigration, it is clear that it has a treaty-based right to do so by supranational means. Such a change is not a mark of a ‘constitutional equilibrium’ or ‘institutional plateau’ and must then be explained.

Liberal Intergovernmentalism (LI) is claimed to have ‘acquired the status of a baseline theory […] which is used as a “first cut”’ when explaining new developments in European integration (Moravcsik and Schimmelfennig, 2009: 67). Indeed, its parsimony and convincing empirical record was enough to motivate its use in this article. However, it becomes even more relevant when its founder, Andrew Moravcsik, claims that there is no legal base for co-operation in the area of migration and, more significantly, that, due to a shortage of functional pressure, there is a slim prospect for its development (2005, 2008). Since a legal base has been created, it becomes pertinent to examine if there indeed were functional pressures or any other explanatory factors. The article then empirically investigates how well LI’s propositions fare with reference to the Swedish case; it demonstrates that Helen Wallace was right when she argued that ‘the asserted hierarchy of decision in three stages – preference formation, negotiation and institutional choice – is too neat and tidy’ (Wallace et al., 1999: 158). In general then, while LI fares well, the article demonstrates that institutional choice is not simply an issue of choosing a means to an end, but rather a preference in itself. Furthermore, it clarifies that Swedish preferences have changed due to unintended processes. To provide a theoretical base for explaining these developments, this article (re-)assembles a typology of the old (neo-)functionalist concept of spillover.

The structure of the article is as follows: the following section introduces LI and develops a typology of different spillovers; the third offers a note on research design, method and material; and the fourth explores the Swedish case; finally, a concluding discussion summarises how well LI fares in light of the empirical findings.

Liberal Intergovernmentalism and spillover

Neofunctionalism (NF) is usually depicted as LI’s main contender for explaining European integration, and much ink has been spilt considering the similarities and differences between them (Moravcsik, 2005; Rosamond, 2000; Schmitter, 2004). Ernst B Haas claimed that LI’s ‘core assumptions are identical with those of NF’ (Haas, 2001, endnote 10), and Philippe Schmitter maintains that if Moravcsik expanded LI to include that states’ preferences also are affected by transnational firms, associations and movements, ‘his approach would be virtually indistinguishable from neofunctionalism’ (Schmitter, 2004, endnote 2). Moravcsik rejects this, and summarises the difference by postulating that NF explains change ‘primarily through endogenous spillover, while LI explains it as a response to exogenous pressure or intended consequences of previous agreements’ (2005, endnote 18); this seems to be a fair description.

While it is claimed that the new theoretical battle line is elsewhere (Pollack, 2001), it is nonetheless easy to agree with Philippe C Schmitter, ‘whose youthful flirtation with neofunctionalism had long since past’, when he states that, when reading recent efforts, the theoretical core ‘often sounded quite familiar’ (Schmitter, 2004: 45). Despite this, it must be explicitly acknowledged that LI is very different from the earlier, (however thoughtful) less sophisticated Intergovernmentalist/Realist approaches (Hoffmann, 1964).

The article’s point of departure is to investigate to what extent LI withstands empirical scrutiny in the Swedish case. Essentially, LI is made up of three theories. The first theory concerns how states’ preferences are formed. Even though LI, like Realism/Intergovernmentalism, views states as the central actors, there is an important difference: LI does not take preferences such as ‘security’ for granted. Rather, LI employs a ‘liberal’ theory for how national preferences are formed, and suggests that such preferences vary according to issue-specific societal interdependence and domestic institutions. For example, when it comes to economic issues it assumes that national preference is derived from a balance between producer interests on the one hand, and taxpayers and those interested in regulation on the other. However, when it comes to immigration it assumes that producers will not be involved. At this early stage then, this article verifies the absence of domestic actors.

LI’s second theory is a bargaining theory for international negotiations. To explain who gets what, LI assumes that asymmetrical interdependence affects the outcome of substantive negotiation. Thus, the actors in least need of an agreement are more able to threaten with non-co-operation the actors with greater need. From this it may be assumed that Sweden, which over the last decades has received a proportionally high number of refugees, will have less clout since it – assuming an interest in spreading the cost – would be in greater need of co-operation.

LI’s third theory is a functional theory of institutional choice. This implies that states, after they have bargained the substantive agreement, create institutions that are appropriate for the agreement to be maintained. To create credible commitments, it may in some cases even be necessary to pool and delegate sovereignty (Moravcsik and Schimmelfennig, 2009: 69ff).

The article shows that while LI in many instances fares well, it also has shortcomings. The research shows that Sweden unintentionally came to change its original goal to preserve decision-making with unanimity. When coming to grips with this empirical development, it seemed necessary to revisit and employ the spillover concept. In this endeavour, it became obvious that there was a general need for a more coherent typology of different spillovers, and this is also introduced below.

Spillover, generally understood as self-reinforcing processes that may take place once an integration process has been initiated, has its background in the functionalist theory that found its most elaborated formulation in David Mitrany’s work A Working Peace System, but in fact the concept goes back to the 19th century (Mitrany, 1943). The concept ‘shouldered most of the burden of explaining change’ within NF (Caporaso and Keeler, 1995: 33). While Ernst B Haas’ early writings suggested that spillover was inevitable (Haas, 1958: 49f), he soon retreated from this proposition (Haas, 1961: 376). Despite attempts in the early 1970s to develop ideas able to account for a stagnating integration process, for example spillback and spillaround (see Lindberg and Scheingold, 1970; Schmitter, 1971), the concept became outdated. The realisation of the internal market in the early 1990s renewed interest in the concept (Tranholm-Mikkelsen, 1991). However, in much prominent scholarship the concept was often omitted even though the idea of spillover was evident (Risse-Kappen, 1996; Sandholtz, 1993; Stone Sweet and Sandholtz, 1998). Unsurprisingly, the idea has also been visible in recent theoretical developments, for example sociological and historical institutionalism (Hall and Taylor, 1996; Pierson, 1996), and the actual concept itself also continues to be frequently applied (Ackrill and Kay, 2010; Macmillan, 2009; Ross, 2011). It is nonetheless evident that a more fine-grained distinction between different types of spillovers is of value.

The fundamental ideas behind the different types of spillovers discerned in this article are all traceable to earlier NF scholarship, but analytical distinctions were largely ignored (essential references are Haas, 1958, 1964; Lindberg, 1963; Lindberg and Scheingold, 1970; Nye, 1971; Schmitter, 1971). While the concept often is associated with NF, it is of paramount importance to state that LI’s rationalistic propositions allow actors to predict, and even intend, that certain actions may result in further integrative measures. Thus, a first cut is the one between intended and unintended spillover. This distinction is important, since it goes to the core of one of the most important discussions with regards to European integration: while LI easily accommodates intended spillover, its rationalistic base cannot handle unintended ones.

Jeppe Tranholm-Mikkelsen (1991) discerned three different forms of spillover: political, cultivated and functional. These are all employed here, but with the important addition of a fourth: social spillover. Political spillover takes place if interest groups, once a policy area has been integrated, turn their lobbying efforts to the regional level, thereby increasing political pressure in favour of further integration (George, 1985: 22f). LI can accommodate that the states, for example, may predict that co-operation within a new area will result in lobbying at the EU level. However, it does rule out that such political pressure may result in further integration than originally intended. Behind the concept of cultivated spillover lies the assumption that supranational actors, primarily the Commission, are relevant since they deliberately link political issues into package deals that make them more acceptable to national governments.¹ Since it suggests that the member states neither can predict what the supranational actors will do, nor restrain them, cultivated spillover can only be unintended. Functional spillover refers to ‘a situation in which a given action, related to a specific goal, creates a situation in which the original goal can be assured only by taking further actions, which in turn create a further condition and a need for more action and so forth’ (Lindberg, 1963: 10). A functional spillover can be both unintended and intended. The latter is in line with LI and, as will be shown, the member states were at an early stage discussing that the realisation of the internal market would necessitate more co-operation with regards to migration. Finally, there is the possibility that during the integration process the actors may experience conflicts between different goals, and/or between means and goals, and thus make new trade-offs that result in an abandonment of their original goals, in other words a social spillover. The issue at stake here is that the integration process may result in states changing their original preferences. While Joseph S Nye’s terminology ‘simple’ learning – adapting the means to attain the same goal – is perfectly in line with LI, ‘complex learning’ (social spillover) – recognition of conflicts between means and goals that result in new preferences – is not (Nye, 1987: 380, drawing on Argyris and Schon, 1978).

Unintended or intended functional spillovers and unintended political, cultivated and social spillovers are all explanatory concepts. However, a distinction that does not seem to have been made earlier is that the spillover concept may as well be used in a descriptive way. As a descriptive concept, it is essential to distinguish between widening (‘sectoral’) and deepening spillover. While the former refers to when integration expands from one sector to another, the latter involves a more far-going delegation and pooling of power (Caporaso and Keeler, 1995: 31).

Finally, it must be acknowledged that it is with hesitation that the concept of spillover is resuscitated. Even though it is common to find it applied in contemporary research, there were good reasons why the major theorists abandoned it. The problem was, and continues to be, that while it may be rather easy to discern a spillover after it has taken place, it is more difficult to specify the conditions under which a spillover will take place. While this may be less surprising with regards to unintended spillovers, it is quite problematic with regards to spillovers that may be, if not intended, at least foreseen. For example, as early as 1985, the Commission claimed that the free movement of persons would demand co-operation on asylum and immigration (1985, points 11 and 55). Despite the impressive developments in terms of treaties, actual co-operation has not developed that much. Nevertheless, while these shortcomings are real and important, until more adequate theoretical constructs are developed, spillover has a place in the conceptual toolbox.

A note on research design, method and material

The choice to study migration at the EU level is rather easily motivated given that a prominent researcher, namely Moravcsik, claims that there is no such institutional base, and that, due to a shortage of functional pressure, there is a slim prospect for its development (Moravcsik, 2005, 2008). As such, a case study of the policy area in itself seems ‘least likely’ to confirm LI’s propositions (Eckstein, 1975), yet such an institutional base has been created. Regarding the choice of Sweden, I was aware of the remarkable change from ardent opponent to supranationalism to its advocate, but could not foresee what theoretical propositions this empirical choice in the end would support. As such, there were many good reasons to turn to LI ‘as an essential first cut explanation’ (Moravcsik and Schimmelfennig, 2009: 67).

The investigation is carried out through process-tracing (George, 1979; George and Bennett, 2005). Consequently, based on LI’s three theories, the research establishes through a careful analysis the causal mechanisms that made Sweden become an advocate of supranationalism within the field of migration. Such an analysis entails clarifying to what extent the claimed justifications were the same as the actual reasons for the action. Several techniques were utilised, for example control for consistency over time and cross-checking of the information given in different contexts. Such information was found via, for example, the shorthand records taken at the Committee on EU Affairs, and completely unrestricted access to the Swedish Ministry of Foreign Affair’s internal documents. Furthermore, interviews were carried out with 10 centrally placed persons, including the State Secretaries and Heads of Negotiations in charge of the Swedish participation at the different Intergovernmental Conferences (IGCs).

The process-tracing revealed that the Swedes were overwhelmingly interested in only one substantive issue: the change from EU policies deciding only ‘minimum standards’ to the current ‘standards’. When it came to the decision-making process, the Swedes essentially discussed only one change: if the Council is to make decisions using QMV (‘supranationalism’) or unanimity (‘intergovernmentalism’). During the analysis, attention was paid to what extent other ‘markers’ of intergovernmentalism or supranationalism (the Commission, the Parliament or the Court) surfaced, but the data suggests that the essential discussion concerned QMV.

Explaining the development of a supranational legal base

The article now discusses the changes that took place with the treaties of Maastricht, Amsterdam, Nice and Lisbon.² It will do so without an in-depth description of the content of the treaties; for this, Steve Peers invaluable and monumental EU Justice and Home Affairs Law (2011) is recommended.

The Treaty of Maastricht

Before the Treaty of Maastricht came into effect in November 1993, the institutions of the EU had no formal role in the area of asylum and immigration. With the exception of two specific issues concerning visas (Art 100 c), this arrangement was maintained in the Maastricht Treaty. Asylum and immigration policy was regarded as matter of ‘common interest’ (art K.1), the Commission shared power with the member states (Art K.3 (2)), the Parliament had quite limited powers (Art K.6) and the Council largely made unanimous decisions (Art K.3 and 4).

Sweden was not a member when the EU was formed but a very unusual report to the European Council’s Maastricht meeting from the ministers responsible for immigration warrants discussing the treaty anyway (Ministers, 1991). Even though the report does not say anything about specific national preferences, it lends itself to a theoretically interesting analysis with regards to the motivations behind the modest steps towards integration taking place at that time.

In the report, the ministers claim that one reason for a common policy was the prospect that freedom of movement for persons would be attained in a few years and this would make co-operation necessary (Ministers, 1991, A II). In theoretical terms, this is a functional spillover that is perfectly in line with LI, since it is foreseen and intended. However, the ministers also argue that ‘harmonization of basic asylum policy is … merely a logical step’ since ‘the ultimate effect of [some initial Conventions] is much greater than was perhaps originally expected’ (Ministers, 1991, A I, italics added). References to unintended spillover do, of course, go against the grain of LI, but, as a whole, the report supports LI’s assumptions. The ministers do rank the arguments in favour of continued integration. Leaving the stated arguments aside, the ministers write that the ‘deeper causes calling for a harmonized immigration policy may be instanced’, namely that ‘the pressure of immigration on most Member States has increased significantly’, and thus ‘the conviction that… a strictly national policy could not provide an adequate response has been consistently gaining ground’ (Ministers, 1991, B I). As such, one can only conclude that the minor steps taken with the Maastricht Treaty are in line with LI’s assumptions that states’ preferences are formed as responses to managing globalisation.

The Treaty of Amsterdam

With regards to asylum and immigration, the Amsterdam Treaty that came into effect in October 1999 implied an abandonment of the use of regular international law and non-binding measures, such as recommendations. While it is easy to describe that any new rules were to be in the form of ‘standard’ Community instruments, such as binding directives, other changes are less easily summarised. Besides the specific arrangements of the United Kingdom, Ireland and Denmark with regards to the integration of the Schengen acquis into the EU frame, there was real complexity regarding the rules on decision-making. Some visa issues were, for example, to be immediately decided with QMV and others with unanimity, which would automatically switch to QMV after five years. However, in general the Council would decide unanimously if all, or remaining parts, of the treaty’s asylum and immigration issues should be subject to QMV (Art 67. TEC).

Andrew Moravcsik and Kalypso Nicolaïdis have given a more general explanation of the process and outcome of negotiating the Treaty of Amsterdam in which they sporadically mention the issue of immigration. They conclude that one of the big actors, Germany, pushed the issue due to asymmetrical interdependence, while the UK and Ireland, which have stronger control over their borders, strongly opposed it (1999: 63). The statistics also reveal that Germany in 1993 received more than 322,000 asylum applications, while the UK received 28,000 and Ireland 90. However, Sweden was even more asymmetrically interdependent than Germany, since in 1993 it received more than 84,000 asylum applications despite having less than a ninth of Germany’s population (UNHCR, 2003, Table 2). Fully in line with LI, Sweden was also interested in greater EU involvement, and the Swedish experience of the early 1990s continued to influence the Swedish position during the coming IGCs. These experiences are, for example, spontaneously brought up in interviews, by people like the Swedish chief negotiator, Gunnar Lund, and the state secretary, Gun-Britt Andersson (Andersson, 2007a; Lund, 2007). However, it is not only Sweden’s greater asymmetrical interdependence, but also its expectation that a realisation of a free movement of persons will necessitate further common policies, that explains why Sweden, already early on in its EU membership, had a strong preference for greater EU involvement (Government Bill 1994/95:19: 406, 1995/96:30: 21).

With regards to interstate bargaining, one may expect that Sweden’s greater asymmetrical interdependence would create an interest in advancing substantive measures. However, the research does not reveal what the Swedish government wanted substantively, that is the exact policies to handle its greater asymmetrical interdependence and the consequences of a free movement of persons. Rather, the Swedish government’s written communication to parliament ahead of the negotiations shows only a concern regarding the choice between supranational or intergovernmental means. The Swedish approach was in general intergovernmental, but the government also stated that it should ‘be considered’ if ‘appropriate parts’ regarding asylum and immigration should imply the involvement of the Commission and a change to community legislation, for example to start using directives and decisions (1995/96:30: 21). Nevertheless, with the latter minor exception, the Social Democratic government entered the Amsterdam negotiations with the aim of maintaining an intergovernmental set-up and, consequently, found itself in ‘the minority that completely rejected the use of QMV within this area’ (MFA 1996-05-20: 2).

A result of the Amsterdam Treaty was that the Council began using QMV a few years after the treaty came into effect. During the negotiations, Sweden was consistently hesitant with regards to QMV since it concerned, as the minister of foreign affairs stated, ‘very sensitive issues on asylum and so on’ (MFA 1997-02-26). Moravcsik and Nicolaïdis’ claim that Germany’s initial push for more QMV was a result of its asymmetrical interdependence does not provide a complete explanation, since Sweden then would have done the same.³ It is evident that the issue of QMV or continued decision-making with unanimity was not, as LI supposes, a question of institutional choice, but, in the Swedish case, a question of a substantive preference.

During the bargaining, the choice of institutional set-up was fundamental. LI’s assumption that governments pool sovereignty ‘in order to “lock in” compliance with particular arrangements or relative influence over future decisions’ (Moravcsik and Nicolaïdis, 1999: 76) seems reasonable. However, if a state does not want to be ‘locked in’, one has to explain why it nonetheless may accept it. What finally made Sweden accept QMV was a break-through with regards to the substance of the rules. The stenographic records taken at the Committee on EU Affairs clearly show that even though Sweden felt the pressure to accept QMV during the IGC, Sweden maintained that QMV could only be accepted if the substantive measures would only concern minimum standards, and thereby allow individual states to pursue more generous policies (Swedish chief negotiator IGC 1996/97:23, statement 16, IGC 1996/97:24, statement 2 and 8, IGC 1996/97:29, statement 9 and 16. Also Government Bill 1997/98:58, part 1: 29). This became the result, and can be understood as a side-payment to Sweden and other like-minded states.

To summarise, the number of asylum seekers to Sweden and, by extension, interest in EU co-operation, is in line with LI’s first assumption that interdependence may explain national preference. However, with regards to bargaining, LI suggests that Sweden’s disproportionate responsibility with regards to asylum seekers and, thereby, its greater asymmetrical interdependence, should make it prone to make concessions. However, it was Sweden that resisted an extension of QMV until it received a side-payment in substantive terms (that the EU was only going to decide on minimum standards). Thus, while the Swedish case shows that asymmetrical interdependence can explain why states’ preferences differ, it also shows that the institutional set-up is not only a choice of means for handling such interdependence, but may also be a national preference.

The Treaty of Nice

Within the field of migration the Treaty of Nice, signed in February 2001, implied only minor changes and simply speeded up some of the successive modifications stipulated in the Treaty of Amsterdam. The Swedish change in position is then not only more extensive but also more surprising. In a governmental instruction to a preparatory meeting before the IGC, it was stated that ‘a general transition to qmv is probably not in the Swedish interest’, but that with regards to asylum and immigration there was an interest in advocating such a transition (MFA 2000-02-11: 10). In later instructions this position was not only repeated but reinforced, since it was stated that such a transition should take place as soon as possible, and be extended to as many issues as possible within the area of asylum and immigration (MFA 2003-03-25: 3, MFA 2004-04-11: 7). What then explains such a complete overhaul of the Swedish position?

In essence, the Swedish decision-makers became convinced that decisions taken by QMV represent ‘[…] probably […] the only realistic model for decision-making when one is to substantiate a common asylum and immigration policy, especially within an enlarged EU’ (Written communication from the Government to the Parliament, 2003/04: 13). Behind this statement was the perception that the demand for unanimity had made it difficult, not to say impossible, to reach the substantive goals the member states earlier had agreed on (MFA 2000-07-03: 38). However, one wonders why Sweden changed its position after the Treaty of Amsterdam. There was an already established understanding that unanimous decision-making decreases efficiency (for example, see Government Bill 1997/98:58: 54). Indeed, it was the actual experience of the day-to-day negotiations, aiming to add actual substance to the Treaty of Amsterdam, that lay behind the Swedish change. In the words of the Director General Ola Henrikson at the Ministry of Justice, the Swedes experienced that negotiations

…went rather slow and that certain states could block solutions that we perceived were acceptable or good. They could push issues in a direction that we perceived they would not have been able to do if we had used qualified majority voting. It was a reaction from the floor, we who were negotiating on a working-group level, that things were very slow. (Interview 28 March 2007; author’s translation)

The frustration that this awoke is vividly illustrated in a memo that, according to the author’s information, was not supposed to reach the Swedish parliament. Concerning the directive on the reception of asylum seekers the memo stated that:

[Sweden] has worked actively to reach a, for us, acceptable solution … Unfortunately, the directive has been very watered down and therefore it ought to be questioned if the member states can be proud of the adoption [of the directive] since we actually have not agreed upon anything. Frankly, we could as well paste together the member states’ different rules on reception and call it an EU-directive. (Memo to the Parliament, 2002; author’s translation)

The memo also concluded that since the negotiations were conducted within the confines of unanimity, the chance for greater changes were very small. Interviews with the political leadership in the Government reveal that it was aware of the negotiators’ frustrations and, even more, with the lack of progress within the area (Andersson, 2007a; Andersson, 2007b). It concluded that it was primarily governments against the more liberal rules that Sweden favoured that used the veto during the negotiations. Furthermore, the Swedes became convinced that the actual negotiation processes were too lax as a result of the demand of unanimity. Thus, when states know that they in the end can veto a proposal, they do not actively take part and, consequently, it is difficult to create a constructive process (Andersson, 2007a).

To conclude, all evidence suggests that the fundamental U-turn Sweden made with regards to institutional choice is explained by their day-to-day experience of negotiations after the Amsterdam Treaty, during which it learned that the intergovernmental venue was insufficient for creating credible commitments. This U-turn is perfectly compatible with LI if one accepts that institutional choice is only a question of selecting an appropriate means. However, the article’s earlier empirical findings show that the choice of institutional set-up may also be a substantive preference. Understood this way, the change conceptually can be understood as a social spillover; in other words, an abandonment of an original goal. Nevertheless, the change was restricted only to the decision-making procedure, and the Swedes still considered it of paramount importance that co-operation was limited to minimum standards (Committee on EU-Affairs: State Secretary Andersson 1999/2000:5, statement. 17; Minister of Immigration Klingvall 2000/01:1, statement. 139). This too was to change in the coming years.

The European Convention and the Treaty of Lisbon

In order to pave the way for a new IGC and a new treaty, in 2001 the European Council decided to convene a convention made up of representatives of the heads of states or governments of the member states, members of national parliaments and of the European Parliament, and Commission representatives. Without dwelling on the complicated ratification process, one can conclude that, concerning asylum and immigration, the European Convention’s suggestion for a ‘Draft Treaty establishing a Constitution for Europe’ (OJ, C169, 18/07/2003) was in essence transferred untouched into the Treaty of Lisbon, which came into effect on 1 December 2009. The implication was that, with the exception of the member states’ continued right to determine the number of third-country nationals seeking work (Art 79(5) TFEU), the EU would develop a common policy on asylum and immigration through supranational means, such as Council decisions with QMV and with the Parliament co-deciding.

The actual asylum policies that the EU will decide on are more or less the same as those enumerated in the Treaty of Amsterdam. However, there is one fundamental difference, namely that the earlier key concept of ‘minimum standards’ is removed. As such, there is no longer a constitutional base for states to pursue more liberal rules with regards to, for example, the reception conditions of applicants for asylum, or the procedures for granting or withdrawing asylum protection status. Given that this was a side payment that Sweden secured during the Amsterdam IGC in exchange for accepting the incremental implementation of QMV, the Swedish case is once again intriguing. During the Nice IGC Sweden, as a result of its experiences of day-to-day negotiations, abandoned its resistance towards QMV, but what can explain Sweden accepting the abandonment of the principle of minimum standards?

In keeping with the importance Swedes attached to ‘minimum standards’, it is evident that this was not an easy change. An indication of this is visible in the fact that, during the European Convention, five out of the six Swedish representatives made a written suggestion that minimum standards should be kept (Hjelm-Wallén et al., 2003). However, during the IGC that followed the European Convention, the Social Democratic government changed its mind and, with the support of the Liberal-Conservative Party (Moderaterna), Sweden came to accept the abandonment of minimum standards.

It is clear that there was an element of negotiation tactics that led Sweden – the Social Democratic government – to change its mind. The Swedes early on discovered that a clear majority of the member states wanted to remove the minimum standards and, based on the perception that a small state cannot push too many issues at an IGC, the decision was taken to drop minimum standards. However, it is highly implausible that Sweden would have changed its mind if it had not also reconsidered the substantive issue itself. When the idea of minimum standards was introduced, the Swedes believed it would imply that Sweden and other (like-minded) states could maintain more liberal standards, while establishing a (higher) baseline for all other states. The Swedes now questioned this assumption, suggesting that minimum standards may rather lead to a ‘race towards the bottom’. Thus, it seemed that in practice other states did not actually maintain more liberal rules, but rather adapted their national legislation towards what were truly ‘minimum’ standards. The change in the Swedish position is succinctly summarised by the Swedish Chief Negotiator, Lars Danielsson:

Before the Treaty of Nice, there grew a general realization that the area of asylum and immigration is one of the policy areas where European co-operation is very beneficial for us. … We see that a continued national asylum and immigration policy only result in asylum seekers moving to the state that has the most liberal rules. And that is not sustainable. That is a reality that gradually has become clear for social democratic politicians and that has partly collided with the principled conviction of intergovernmental co-operation. But in the end intergovernmental co-operation has given way. (2007, author’s translation; this account is corroborated by Andersson, 2007b; Petersson, 2007)

In theoretical terms, such a change on a substantive national preference is appropriately conceptualised as an unintended, social spillover.

Concluding discussion

The article’s empirical findings suggest that employing LI as a baseline theory was not enough to conceptualise the fundamental change in Swedish positions with regards to supranationalism and migration. Further, the data analysis suggests that, even though there remain problems with the concept of spillover, for example, in specifying when it will take place, it clearly remains of value. While working with the concept, it was deemed necessary to create a more coherent typology of different spillovers, indicating that the concept of spillover has both descriptive and explanatory value. With regards to descriptive spillover, it seems valuable to differentiate between widening and deepening spillovers, but concerning explanatory spillovers, more options became visible: there are unintended or intended functional spillovers, as well as unintended political, cultivated and social spillovers. In the article, it is clear that while LI easily accommodates intended, functional spillovers, it has problems with the identified unintended social spillover that resulted in a widened as well as deepened spillover. The empirical findings do not suggest that political or cultivated spillover have taken place.

Generally, LI is completely in line with the conclusion that the EU’s supranational asylum and immigration policies developed as a result of exogenous pressure – increased number of asylum seekers – and as an intended, functional spillover – the perceived need to compensate for free movement of persons within the EU. However, LI focuses more on individual states and, as a result, weaknesses appear. LI proposes that ‘Governments debate institutions only once they strike a substantive bargain…’ (Moravcsik, 1998: 21f), but this article’s detailed study of Sweden shows that this is not the case. Further, an overview of all member states’ positions reinforces that the issue of institutional choice may be the most important one for states. A report before the Amsterdam IGC shows that, with the exception of Italy, states almost exclusively paid attention to issues such as communitarisation and QMV, but not to substantive content (Hix and Niessen, 1996). As such, while all member states had a preference for greater EU involvement, the issue at stake was the institutional set-up and not actual substantive policies. The obvious implication is, then, that institutional choice may also be a national preference, and not necessarily something debated after the substantive bargaining process has been completed.

Theoretically, LI does not consider ‘ideological’ stances, such as certain states being more or less reluctant to engage in supranationalism. However, on the one hand, Moravcsik acknowledges that there have been instances when ideological motivations have altered the national position (1998: 474, 488). On the other hand, Moravcsik gives ideology a ‘secondary yet significant role’ only in ‘situations in which the substantive risks and substantive consequences of decisions are modest’ (1998: 486). Generally, it seems difficult to claim that the creation of a constitutional base for migration concerns represents a situation with modest consequences. Nevertheless, the initial Swedish preference to continue to make decisions with unanimity was ideologically based, in line with its traditional intergovernmental instinct, and more significant than its reception of an asymmetrical number of asylum seekers and concerns about the free movement of persons. Admittedly, the case at hand is limited: it concerns one member state’s preferences in one policy area, but the clear indication that the other member states also discussed institutional choice primarily as a substantive preference suggests something interesting. It may be that LI’s claim that institutional set-up is only a means to an end is more accurate when it concerns economic matters. However, when EU co-operation touches upon policy areas where ‘sovereignty is nowhere more absolute’ (Arendt, 1973: 278), ideology comes to play. Inevitably, Hoffmann’s old distinction between ‘high’ and ‘low’ politics comes to mind (1964: 1274f.). Areas of high politics imply ‘sovereignty costs’ being at their highest, since international co-operation then touches ‘on the relations between a state and its citizens or territory, the traditional hallmarks of (Westphalian) sovereignty’ (Abbott and Snidal, 2000: 437). With regards to future research it may well be worth exploring if low sovereignty costs are a scope condition for an otherwise most impressive baseline theory, LI, which fares less well when the issue entails high sovereignty costs.

LI’s second theory suggests that states least in need of an agreement are more able to threaten other states with non-co-operation. Considering Sweden’s disproportionate responsibility-taking with regards to asylum seekers, LI then predicts that it should be in urgent need of co-operation and, thus, make concessions. However, at the Amsterdam IGC it was Sweden that resisted an extension of QMV until it received a side-payment in substantive terms – that the EU was only to decide on minimum standards.

LI’s third theory suggests that states’ institutional choice to pool sovereignty is a result of a need for credible commitments. As indicated above, the Swedish position at the Amsterdam IGC shows that institutional choice is something more than merely choosing a means to an end. Thus, the institutional choice can be the preference. Nevertheless, the Swedish U-turn with regards to QMV at Nice shows that LI also has something important to offer in the case of asylum and immigration. In the day-to-day work of filling the Amsterdam Treaty with actual substance, the new EU member, Sweden, learned the hard way how important it was to have credible commitments. This lesson was so significant that it influenced Sweden into changing its position with regards to institutional set-up, and to ultimately advocate QMV. One may argue that this development shows that LI is correct and that asymmetrical interdependence is more important than institutional choice and, as such, it is to be expected that the latter yields. However, the empirical evidence strongly suggests that at the Amsterdam IGC the issue of maintained unanimity was so crucial that Sweden’s asymmetrical interdependence would have come second if it had not been for the proposal that the EU would only decide on minimum standards. Thus, the Swedish change with regards to QMV, from hard resistance to QMV to advocating it, can only be described as a social spillover. This conclusion applies even more to the Swedish change during the latest Treaty revision when Sweden altered its position regarding minimum standards.

Even though the discussed social spillover is, on the one hand, anathema to LI, it does, on the other hand, validate Moravcsik’s empirical claim that ‘The integration process did not supersede or circumvent the political will of national leaders; it reflected their will’ (1998: 4; italics in original). While this statement is correct, an important point to note is that the will of national leaders can change. The Swedish case shows that such a change may take place not only with regards to institutional choice – which the article shows to be a national preference – but also concerning substantive policies. The Swedish journey from resistance to QMV to advocating it, and the removal of ‘minimum standards’, illustrates that previous decisions can result in states abandoning their original preferences.

Footnotes

Acknowledgements

I greatly appreciate the thoughtful comments of the three anonymous reviewers. Thanks also to Lia Antoniou for the work that she has put in during the process of finishing this article.

Funding

This work was supported by the Swedish Research Council and Swedish Institute of International Affairs.

Notes

Author biography

Hans E Andersson is the Chief Administrative Officer and a Senior Lecturer in Political Science at Södertörn University, Sweden. His research interests include international politics, and immigration, integration and asylum policies (with an emphasis on the cooperation within the EU). His publications include Rights and Citizenship in Transition (with Lia Antoniou, forthcoming, Palgrave Macmillan) and articles in Ethnic and Racial Studies, International Relations and The International Journal of Children’s Rights.

References

Abbott

Snidal

(2000) Hard and soft law in international governance. International Organization 54: 421–456.

Ackrill

Kay

(2010) Multiple streams in EU policy-making: The case of the 2005 sugar reform. Journal of European Public Policy 18: 72–89.

Andersson

G-B

(2007a) State Secretary for Immigration 1996-2002. Interview 13 June.

Andersson

(2007b) Political Advisor at Prime Minister’s Office. Interview 3 April.

Andersson

(2008) Överstatlig flyktingpolitik. Staters samarbeten inom suveränitetskänsliga områden. Umeå: Boréa Bokförlag.

Arendt

(1973) The Origins of Totalitarianism. New York: Harcourt Brace Jovanovich.

Argyris

Schon

(1978) Organizational Learning: A Theory of Action Perspective. Reading, MA: Addison-Wesley.

Brochmann

(1996) European Integration and Immigration from Third Countries. Oslo: Scandinavian University Press.

Caporaso

Keeler

(1995) The European Union and regional integration theory. In: Rhodes

Mazey

(eds) State of the European Union, Vol. 3: Building a European Polity? Boulder, CO: Lynne Rienner Publishers Inc., pp. 29–62.

10.

Collinson

(1993) Beyond Borders: West European Migration Policy towards the 21st Century. London: Royal Institute of International Affairs.

11.

Commission (1985) Completing the internal market. COM (85), 310, Final, 14 June. Brussels.

12.

Danielsson

(2007) State Secretary for EU Policy from 1999 and the Prime Minister’s State Secretary 2002-2006. Sweden’s chief negotiator at the IGC preparing the Constitutional Treaty. Interview 30 March.

13.

Donnan

Wilson

(1999) Borders: Frontiers of Identity, Nation and State. Oxford: Berg.

14.

Eckstein

(1975) Case study and theory in political science. In: Greenstein

Polsby

(eds) Handbook of Political Science, Vol. 7: Strategies of Inquiry. Reading, MA: Addison-Wesley, pp. 94–137.

15.

Eurostat (2011) Residence permits issued to non-EU citizens in 2009. Statistics in focus 43/2011. European commission. Luxembourg: Eurostat.

16.

George

(1979) Case studies and theory development: The method of structured, focused comparison. In: Lauren

(ed.) Diplomacy: New Approaches in History, Theory, and Policy. New York: Free Press, pp. 43–68.

17.

George

Bennett

(2005) Case Studies and Theory Development in the Social Sciences. Cambridge, MA: MIT Press.

18.

George

(1985) Politics and Policy in the European Community. Oxford: Clarendon Press.

19.

Haas

(1958) The Uniting of Europe: Political, Social and Economical Forces 1950–1957. London: Steven & Sons Limited.

20.

Haas

(1961) International integration: The European and the universal process. International Organization 15: 366–392.

21.

Haas

(1964) Beyond the Nation-State. Stanford, CA: Stanford University Press.

22.

Haas

(2001) Does constructivism subsume neo-functionalism? In: Christiansen

Jørgensen

Wiener

(eds) The Social Construction of Europe. London: Sage, pp. 22–31.

23.

Hall

Taylor

(1996) Political science and the three new institutionalisms. Political Studies 44: 936–957.

24.

Hix

Niessen

(1996) Reconsidering European Migration Policies: The 1996 Intergovernmental Conference and the Reform of the Maastricht Treaty. Brussels: Migration Policy Group.

25.

Hjelm-Wallén

. (2003) Suggestion for amendment of Article 11. Available at: http://european-convention.europa.eu/

26.

Hoffmann

(1964) Europe’s identity crisis: Between the past and America. Daedalus 93: 1244–1297.

27.

Intergovernmental Conference (IGC) Stenographic records, Committee on EU-affairs, Swedish Parliament.

28.

Johansson

(2002) Avslutning: Sverige som EU-medlem. In: Johansson

(ed.) Sverige i EU, 2nd ed. Stockholm: SNS Förlag, pp. 290–297.

29.

Lindberg

(1963) The Political Dynamics of European Economic Integration. Stanford, CA: Stanford University Press.

30.

Lindberg

Scheingold

(1970) Europe’s Would-Be Polity: Patterns of Change in the European Community. Englewood Cliffs, NJ: Prentice Hall.

31.

Lund

(2007) State Secretary for EU Policy 1994-1999, Sweden’s chief negotiator at the Amsterdam and Nice IGC. Phone interview 7 November.

32.

Macmillan

(2009) The application of neofunctionalism to the enlargement process: The case of Turkey. Journal of Common Market Studies 47: 789–809.

33.

Memo to the Parliament (2002) Attachment to Committee on EU- Affairs, 19 April.

34.

MFA – Documents from Ministry of Foreign Affair’s archive. Dossier HP 79/A/EU.

35.

Miljan

(1977) The Reluctant Europeans: The Attitudes of the Nordic Countries towards European Integration. London: C. Hurst & Company.

36.

Ministers

(1991) Report for the Ministers Responsible for Immigration to the European Council Meeting in Maastricht on Immigration and Asylum Policy, SN 4038/91 (WGI 930) 3 December 1991.

37.

Mitrany

(1943) A Working Peace System: An Argument for the Functional Development of International Organization. London: Royal Institute of International Affairs.

38.

Moravcsik

(1998) The Choice for Europe: Social Purpose and State Power from Messina to Maastricht. Ithaca, NY: Cornell University Press.

39.

Moravcsik

(2005) The European constitutional compromise and the neofunctionalist legacy. Journal of European Public Policy 12: 349–386.

40.

Moravcsik

(2008) The European constitutional settlement. World Economy 31: 158–183.

41.

Moravcsik

Nicolaïdis

(1999) Explaining the Treaty of Amsterdam: interests, influence, institutions. Journal of Common Market Studies 37: 59–85.

42.

Moravcsik

Schimmelfennig

(2009) Liberal intergovernmentalism. In: Wiener

Diez

(eds) European Integration Theory. Oxford: Oxford University Press, pp. 67–87.

43.

Niemann

(2008) Dynamics and countervailing pressures of visa, asylum and immigration policy Treaty revision: Explaining change and stagnation from the Amsterdam IGC to the IGC of 2003–04. Journal of Common Market Studies 46: 559–591.

44.

Nye

Jr (1971) Comparing common markets: A revised neo-functionalist model. In: Lindberg

Scheingold

(eds) Regional Integration: Theory and Research. Cambridge, MA: Harvard University Press, pp. 192–231.

45.

Nye

Jr (1987) Nuclear learning and U.S.-Soviet security regimes. International Organization 41: 371–402.

46.

OJ – Official Journal of the European Union.

47.

Peers

(2011) EU Justice and Home Affairs Law. Oxford: Oxford University Press.

48.

Petersson

S-O

(2007) Deputy head of negotiation at the Amsterdam IGC, the Government’s Alternate at the European Convention and Sweden’s chief negotiator at the IGC preparing the Lisbon Treaty. Phone interview 7 June.

49.

Pierson

(1996) The path to European integration. Comparative Political Studies 29: 123–163.

50.

Pollack

(2001) International relations theory and European integration. Journal of Common Market Studies 39: 221–244.

51.

Risse-Kappen

(1996) Exploring the nature of the beast: international relations. Journal of Common Market Studies 34: 53–80.

52.

Rosamond

(2000) Theories of European Integration. New York: St. Martin’s Press.

53.

Ross

(2011) The revenge of neglected issues: EU founders and social policy. French Politics, Culture and Society 29: 90–104.

54.

Sandholtz

(1993) Choosing union: monetary politics and Maastricht. International Organization 47: 1–39.

55.

Schmitter

(1971) A revised theory of European integration. In: Lindberg

Scheingold

(eds) Regional Integration: Theory and Research. Cambridge, MA: Harvard University Press, pp. 232–265.

56.

Schmitter

(2004) Neo-neofunctionalism. In: Wiener

Diez

(eds) European Integration Theory. Oxford: Oxford University Press, pp. 45–74.

57.

SOU (1936) Official Report from the Swedish Government, pp. 53–57.

58.

Stone Sweet

Sandholtz

(1998) Integration, supranational governance, and the institutionalization of the European polity. In: Sandholtz

Stone Sweet

(eds) European Integration and Supranational Governance. Oxford: Oxford University Press, pp. 1–26.

59.

Tallberg

(2001) Inledning - Sveriges ordförandeskap i EU 2001. In: Tallberg

(ed.) När Europa kom till Sverige. Stockholm: SNS Förlag, pp. 9–32.

60.

Tranholm-Mikkelsen

(1991) Neofunctionalism: Obstinate or obsolete? Millennium: Journal of International Studies 20: 1–22.

61.

UNHCR (2003) Asylum Levels and Trends: Europe and Non-European Industrialized Countries, 2003, http://www.unhcr.org/statistics

62.

Wallace

Caporaso

Scharpf

. (1999) Review section symposium. The choice for Europe: social purpose and state power from Messina to Maastricht. Journal of European Public Policy 6: 155–179.