Banking union through the back door? How European banking union affects Sweden and the Baltic States

Abstract Swedish decision-makers opted out of the European banking union (EBU) despite the large cross-border presence of Swedish banks in Estonia, Latvia, and Lithuania. The three Baltic states, on the other hand, have already joined the eurozone and are part of the EBU. This article identifies three important domestic considerations that have shaped Sweden’s position. Firstly, Swedish decision-makers were concerned that member states outside the eurozone would not fully participate in EBU decision-making. Secondly, they were reluctant to pay for the recapitalisation or resolution of distressed non-Swedish banks in other EU countries. Thirdly, Sweden preferred to retain regulatory autonomy in crisis management. The article relates Sweden’s position to the overall cautious approach of other non-eurozone members such as the UK and Denmark. Nevertheless, it highlights the enhanced role of the European Central Bank (ECB) in banking supervision not only for eurozone insiders such as Estonia, Latvia, and Lithuania but also for member states outside the Eurozone such as Sweden.

One of the important rationales for setting up the European banking union (EBU) was to reduce uncertainty in cross-border crisis management and make burden-sharing agreements for bank recapitalisations and bail-outs more transparent. Schoenmaker and Siegmann (2013: 4) have argued that among non-eurozone countries, the UK and Sweden stood to have the biggest efficiency gains from the introduction of EBU. Yet, in Sweden, the economic rationale for joining the EBU was trumped by sovereignty concerns that we seek to unpack in this article. Estonia, Latvia, and Lithuania have already joined the eurozone and, consequently, participate in the EBU. Sweden, by contrast, has remained outside the EBU. It has expressed concerns about being outvoted in the Single Supervisory Mechanism (SSM) decision-making framework, while being liable for bank recapitalisations in other European Union member states.
Sweden and the Baltic states have made different choices regarding EBU, but these countries continue to share extensive links in banking supervision and crisis management. Swedish-based banking groups such as Swedbank and SEB have asserted that they view the entire Nordic-Baltic region as their extended 'home' market. Indeed, in the aftermath of the 2008 global financial crisis, Swedish banking groups delivered on their commitments to stabilise the entire regional market by providing their newly acquired Baltic subsidiaries with sizeable capital injections (Aarma and Dubauskas 2012;Bohle 2013;Kattel and Raudla 2013).
This article aims to make a contribution to the growing literature on the politics of EU banking sector supervision by examining the development of the tightly integrated Nordic-Baltic regional banking system. We investigate the tension caused by the different paths taken by Sweden and the Baltic states regarding membership in the eurozone and EBU. In particular, we seek to explain Sweden's decision not to join the EBU, while the three Baltic states did so, and the consequences of that decision for the Nordic-Baltic regional banking system. In line with other contributions to this volume (Howarth and Quaglia 2016;Schimmelfennig 2016), our analysis of Sweden's position speaks to the overall cautious approach of other non-eurozone members such as the UK and Denmark.
This article follows an explaining outcome process tracing research design (Beach and Pedersen 2013), focusing on the position of Sweden on EBU. The Nordic-Baltic region constitutes our case study, being an example of one the most densely integrated and effective cross-border banking regions in the EU. Within the Nordic-Baltic area, we conduct a paired comparison (Tarrow 2010) analysis of the reasons for the diverging choices of Sweden and the three Baltic states with regard to joining the EBU. Despite Sweden's formal decision to remain outside EBU, we find evidence that the implementation of EBU in the three Baltic states significantly affects the decision-making scope of home supervisors. In particular, we highlight the enhanced role of the European Central Bank (ECB) in banking supervision not only for eurozone insiders such as Estonia, Latvia, and Lithuania but also for member states outside the eurozone such as Sweden. Methodologically, we apply process tracing to identify the causal mechanisms that link domestic factors to policy preferences. We draw on primary and secondary sources such as central bank and supervisory agencies' policy briefs, research notes by the Swedish and Baltic authorities, central bankers' speeches, relevant press coverage, and semi-structured interviews with decision-makers.
This article is structured as follows: we first introduce our analytical framework regarding the factors that we expect to influence countries' preferences about EBU membership. The following section contextualises the importance of supervisory cooperation in the region. Next, we analyse Sweden's decision to opt out of the EBU in light of our analytical framework. Furthermore, we show that the enhanced role of the ECB in banking supervision applies to Swedish cross-border banks despite the opt-out. Lastly, the conclusion summarises the main findings of our analysis.

Analytical framework
We adopt a rationalist framework of analysis, aiming to explain why Sweden opted out of the European banking union. We build on findings in the literature that loss of sovereignty and distributional consequences have been a major concern for EU member states during EBU negotiations (Epstein and Rhodes 2016;Quaglia 2013, 2014). We evaluate Swedish decision-makers' preferences based on primary sources such as statements by key participants in the negotiations and secondary sources such as reports by the ECB and IMF. The three main factors in our explanation refer to: (1) participation in EBU decision-making (Hennessy 2014;Howarth and Quaglia 2014); (2) distributive consequences of the Single Resolution Mechanism (SRM) (Howarth and Quaglia 2014;Véron 2012); and (3) continuation of the maximum harmonisation approach and elimination of national discretions (Epstein 2014;McPhilemy 2014;Spendzharova 2014).
To elaborate on each of these factors in turn: firstly, there is a crucial asymmetry in the status quo pre-EBU for Sweden compared to the Baltic states. The Baltic states all became part of the eurozone with Estonia as front-runner, whereas Sweden has remained outside the eurozone. Sweden is home to large internationalised banks and its home regulators enjoyed significant decision-making power vis-à-vis the host regulators in the Baltic states. However, the decision-making procedures in the new EBU institutions such as the SSM and SRM are more advantageous for member states inside the eurozone. We build on Schimmelfennig's (2016) argument that Sweden's opt-out from the eurozone has produced pressures for further differentiation at the time of deciding whether to join EBU or not. The Baltic states fully participate in EBU decision-making as eurozone members. By contrast, non-eurozone countries such as Sweden, Denmark, and the UK are at a disadvantage even after securing a compromise in the EBU voting modalities (Howarth and Quaglia 2014). Hennessy's (2014) analysis confirms that a key disagreement in the negotiation of the EBU's SSM concerned the dividing line between eurozone insiders and outsiders. Eurozone outsiders such as the UK, Sweden, and Denmark expressed concerns that 'the ECB might become dominant in setting technical rules when it takes over supervision responsibilities' (Hennessy 2014: 163). Consequently, these countries sought protection in the voting modalities of the European Banking Authority (EBA) in order to preserve some decision-making sway. The agreement reached in December 2012 accommodated the UK's demand for a 'double majority' principle, ensuring that EBA decisions are approved by at least a plurality of eurozone outsiders. The compromise was considered a 'very good deal' (Barker and Spiegel 2013), but both UK and Swedish decision-makers maintained that they would not join the EBU in the foreseeable future (Hennessy 2014: 163).
Secondly, the distributive consequences of the EBU's Single Resolution Mechanism are a more sensitive issue for Sweden due to its structural position as a home country to large internationalised banks. On the one hand, Schoenmaker and Siegmann (2013) have argued that the improved EU resolution regime for large international banking groups would be advantageous for both home and host countries (see also De Haas and van Lelyveld, 2014). Furthermore, the supranational approach would help tackle the problematic link between banks and sovereigns (Alter and Schüler 2012). On the other hand, however, the SRM gives home countries less control over recapitalisation decisions, compared to the status quo before EBU (see also Howarth and Quaglia 2014).
The SRM establishes a framework to deal with distressed banks in the EU, together with the Bank Recovery and Resolution Directive (BRRD) (Coeuré 2013;Howarth and Quaglia 2014: 125). Negotiations dragged on for months because Germany pushed for clear bail-in procedures to mitigate moral hazard, while France demanded more flexibility and national discretion (Epstein and Rhodes 2016;Hennessy 2014). European Commission officials feared that giving governments too much flexibility in bail-in regulations would allow wealthy countries to bail out failing banks; only countries with limited resources would impose haircuts on creditors. In turn, this would reinforce the damaging feedback loop between weak sovereigns and weak banks (Véron 2012). Under the 2013 compromise, the SRM bail-in provisions can force shareholders, bondholders, and depositors to bear some of the cost of bank failures. Furthermore, the SRM entails debt mutualisation -member states may be liable for bank rescue operations outside their borders. For Sweden, this has been controversial. Sweden complies with BRRD provisions, but joining the SRM voluntarily in addition to BRRD may entail paying for bank bail-outs of non-Swedish banks.
Thirdly, the ability to preserve national regulatory autonomy is another important consideration when states decide whether to join the EBU. In this regard, the maximum harmonisation approach introduced in the most recent amendment of the EU's Capital Requirements Directive (CRD IV) and maintained in the EBU has been more problematic for Sweden than for the Baltic states. Developing and enforcing the single rulebook has been an important priority for the European Commission and cross-border financial firms since the early 2000s (European Commission 2013a, 2013b). Such an approach curtails the discretions available to national banking supervisors. It aims to create legal clarity and coherence, ensure a level playing field, and foster supervisory convergence across member states. The European Banking Authority and the Single Supervisory Mechanism play a central role in implementing and monitoring the adopted set of harmonised rules.
Yet the maximum harmonisation approach has been a sensitive issue for member states such as the UK, which have consistently pursued more flexibility and national discretion. Sweden has also been in favour of maintaining national regulatory autonomy, especially in crisis management situations when the stability of the domestic banking system is threatened (Spendzharova 2014). After presenting our analytical framework, we now explain how Sweden gained a central role in the tightly integrated Nordic-Baltic banking system.

Enhancing supervisory cooperation in the Nordic-Baltic region
From the end of the 1990s, Swedish banks strategically enhanced their presence in the financial sectors of the three Baltic states. This section shows that complementary to the expanding operations of Swedish-based banks, Swedish bank supervisors were actively engaged in safeguarding financial stability in Latvia, Estonia, and Lithuania. In the course of the 2000s, Swedish-based banks gained a dominant role in the Baltic financial markets (Riksrevisionen 2011: 19). As shown in Figure 1, collectively, Swedish-based banks held more than 80 per cent of total bank assets in Estonia, 60 per cent in Lithuania, and 50 per cent in Latvia. By 2010, Swedish banks had become the major source of credit in the Baltic economies, with combined lending of 400 billion Swedish krona (approximately €45 billion) (Ingves 2010).
The economic turmoil in the Baltic states in the aftermath of the 2008 global financial crisis posed challenges for Swedish banks, central bankers, and bank supervisors. In 2009, Swedish banks incurred considerable losses: 60 per cent of Swedbank's and 75 per cent of SEB's total losses stemmed from their operations in the Baltics (Ingves 2010). Considering the size of the Swedish economy and the strength of its financial system, the losses could be absorbed by the Swedish parent banks. At the same time, as a small and highly internationalised open economy, Sweden strives to maintain investor confidence. The Central Bank of Sweden, Riksbank, was prepared to provide liquidity to the larger Baltic region to ensure regional financial stability (Personal interview, Göran Lind 2012). In order to help contain the economic crisis and rising levels of non-performing loans in the Baltic states, the Swedish authorities cooperated closely with the governments of the three Baltic states, the EU, and the IMF.
The turmoil in the banking sector was most pronounced in Latvia. Among the three Baltic states, Latvia was the only country that required a stand-by agreement with the IMF (Aslund 2011;Engström 2009). As early as August 2008, the country experienced a run on its currency and diminished euro holdings. As the problems continued to escalate, by November 2008, the Latvian government sought balance of payment support from the IMF, the EU, and the Nordic countries. In addition, the governor of the Central Bank of Latvia, Ilmar Rimsevics, approached the Riksbank and requested a currency swap agreement (Personal interview, Göran Lind 2012). The Riksbank was positive about such a swap agreement on two conditions: a binding agreement with the IMF and a transparent crisis management plan. Initially, the Riksbank extended a €500 million swap line (IMF 2010). Subsequently, it provided bridging loans and liquidity to the Latvian financial system, while the country negotiated with the IMF and EU. In December 2008, Latvia signed an agreement with the IMF, the EU, and neighbouring countries. As a result, €7.5 billion were made available to Latvia to meet the external obligations of the troubled Parex Bank, stabilise the financial system, finance the budget deficit, and support the currency peg to the euro. Sweden's contribution to this stabilisation package amounted to €1 billion (Personal interview, Göran Lind 2012).
The Swedish authorities were also committed to safeguarding financial stability in Estonia. In February 2009, the Riksbank offered an agreement to the Central Bank of Estonia for short-term currency support (Ingves 2010). Sweden offered to assume responsibility for 80 per cent of the losses of Swedish-based banks operating in Estonia, while Estonia would bear the remaining 20 per cent (Personal interview, Göran Lind 2012). However, this solution jeopardised Estonia's commitment to maintaining the peg of its currency to the euro and was not put in motion. While Sweden played an active role in maintaining financial stability in the region, we should not underestimate the role of citizens and the relatively low level of social protest in the Baltics despite the implementation of austerity measures. As Bohle (2011) has shown, the willingness of Estonian citizens, in particular, to cope with wage cuts, rising unemployment, and public sector layoffs gave Estonian decision-makers significant leeway in dealing with the effects of the 2008 global financial crisis. The Swedish involvement in Lithuania was limited, given the preference of the Lithuanian government to deal with the impact of the global financial crisis by itself rather than seek external funding (IMF 2010). Even though the Lithuanian government followed pro-cyclical fiscal policy during the economic boom period of 2004-2008, it switched to rapid fiscal consolidation in the aftermath of the global financial crisis (Vilpišauskas et al. 2014). The Central Bank of Lithuania reduced its bank reserve requirements from 6 to 4 per cent and the government increased the coverage of deposit guarantees to €100,000, which was sufficient to calm down depositors and investors. All in all, the Lithuanian banking system was able to withstand the effects of the global financial crisis (Kattel and Raudla 2013). Thus, the Swedish authorities played a limited role in crisis management.
In addition to taking a pro-active role in crisis management efforts in Latvia and Estonia after 2008, the Swedish authorities actively pursued measures to enhance regional cooperation in cross-border banking supervision. In August 2010, the Ministries of Finance, Central Banks, and Financial Supervision Authorities of Denmark, Estonia, Finland, Iceland, Latvia, Lithuania, Norway and Sweden signed a Nordic-Baltic Memorandum of Understanding (MoU) on financial stability, crisis management, and crisis resolution (EBRD 2012: 55). Sweden played a leading role in setting up this initiative. The memorandum paved the way for setting up a permanent regional body, the Nordic-Baltic Cross-Border Stability Group (NBSG), which meets regularly to discuss financial stability issues and has been a hub for regional cooperation during financial crises. The NBSG has also been instrumental in developing ex ante burden-sharing agreements among the participating countries.

Analysing Sweden's decision to opt out of the European banking union
Did Sweden's successful involvement in crisis management in the Baltic region in the aftermath of the global financial crisis influence its decision to opt out of the EBU? As Schoenmaker and Siegmann (2013) have argued, Sweden stood to achieve economic efficiency gains from joining the EBU. Estonia and Latvia had already become eurozone members in 2011 and 2014, respectively, followed by Lithuania in 2015. Moreover, Swedish decision-makers had signalled interest in greater supervisory coordination in Europe even before the 2008 global financial crisis. In 2006, Riskbank governor, Stefan Ingves, proposed the idea to establish a pan-European Organisation for Financial Supervision (EOFS), which would supervise the 40 largest cross-border banking groups in the EU. Aware of the increasing role of cross-border banking, Ingves stressed the need for an institutional body that would be more proficient and reliable in gathering information, overseeing activities of cross-border banks, and making complex risk assessments Ingves 2007). Within the EBU, the SSM fulfils a similar function to the organisation proposed by Ingves. Thus, Sweden's general commitment to greater coordination at the European and international level makes the decision to opt out of the EBU somewhat puzzling.
In the analytical section we put forward three factors that could potentially explain Sweden's position on EBU derived from the literature on the politics of EU banking supervision: ability to fully participate in EBU decision-making; distributive consequences of the SRM; and continuation of the maximum harmonisation approach introduced in the Capital Requirements Directive. To elaborate on each factor in turn, firstly, as a non-eurozone member, Sweden would not be able to fully participate in EBU decision-making. Secondly, the Swedish authorities were apprehensive about a systematic transfer of resources from Sweden to other countries, which would be a liability in the eyes of Swedish taxpayers. Thirdly, the Swedish authorities preferred more national regulatory autonomy in crisis management.
Regarding the first factor in our analysis, as highlighted by Hennessy, eurozone outsiders such as Britain, Sweden, and Denmark expressed concerns that 'the ECB might become dominant in setting technical rules when it takes over supervision responsibilities' (Hennessy 2014: 163). Prior to the introduction of EBU, decision-making powers in bank supervision resided mostly with the Swedish government and supervisory authorities. With the ECB in charge of supervising systemically important banks in the EU, Sweden would not have comparable influence in EBU decision-making.
The Swedish authorities acknowledged the benefits of greater supervisory cooperation at the EU level in order to cope with the ongoing economic and financial problems in Europe. In principle, they supported EBU. However, full voting rights and an equal say in EBU decision-making were of paramount importance. The contentious nature of the issue was clear during the meeting of EU finance ministers in Cyprus in September 2012. Dissatisfied with the EBU proposals discussed in the meeting, Sweden's Finance Minister Anders Borg explained that 'they could particularly not accept the supervision based on the ECB, where they cannot become members without joining the euro' (Financial Times 2012a). Sweden's Prime Minister Fredrik Reinfeldt voiced similar concerns when commenting on the EBU: 'What is in it for us? If we are without influence, that will never fly. I don't think any of the [eurozone] outs will accept that' (Financial Times 2012b).
Even though a partial compromise gave any EU member state that decides to join the EBU equal voting rights in the Supervisory Board of the SSM, Swedish decision-makers were not fully convinced. The Supervisory Board of the SSM is subordinate to the ECB's Governing Council, which holds the ultimate decision-making authority and in which the EBU opt-ins are not represented (CEPS 2014: 27). Member states outside the eurozone deciding to opt in to the EBU are also at a disadvantage because they cannot participate in the liquidity provision operations of ECB unless they have substantial operations and collateral in the eurozone (CEPS 2014: 27). Given these serious reservations, the Swedish government clarified that it would not seek to join EBU in the foreseeable future. The decision was based on a broad domestic consultation. In October 2013, the Swedish parliament's Finance Committee declared that the EBU legislative proposals contradict the principles of subsidiarity and proportionality. Consequently, the Swedish parliament used the Lisbon Treaty 'yellow card' procedure to signal its concerns to the European legislative institutions (European Parliament 2013: 3). It warned that the EU institutions would gain disproportionate influence over the potential allocation of Swedish taxpayer revenue for bank recapitalisation in other member states.
A second concern informing the Swedish position to opt out of the EBU was based on Sweden's potential obligation to pay for the recapitalisation or resolution of weaker European banks in the EBU. The Swedish government and parliament identified considerable fiscal and accountability risks associated with the obligation to participate in the SRM. They were particularly worried about the funding of bank resolution operations and deposit guarantee schemes. The Swedish financial system was, by and large, safe and the banks were well capitalised (Ekici et al. 2009;Goodhart and Rochet 2011). The Swedish domestic credit market did not experience boom and bust dynamics of the type seen in the USA or the Baltic states (Bayram 2014;Englund 2011). The losses of the Swedish subsidiaries in the Baltics were absorbed by the parent banks in 2010 and 2011 without spill-over effects on the Swedish domestic banking system. Furthermore, both Swedish banks and the government were in a better financial position to withstand the effects of the global financial crisis compared to other EU member states (Wehner 2007). Drawing on their crisis management experience of the 1990s, the Swedish authorities had altered the budgetary process, paying particular attention to running budget surpluses that could be used to mitigate recessions (Jonung 2008). Having invested in building a more resilient domestic financial system, Swedish decision-makers were not prepared to join a supranational resolution mechanism that might entail a systematic transfer of resources from Sweden to other member states. Such a transfer would be a clear electoral liability, particularly taking into account that as a country outside the eurozone Sweden would not have a full say in EBU decision-making.
Lastly, we argue that the importance of national regulatory autonomy in crisis management is a third factor explaining Sweden's decision not to join EBU. As Howarth and Quaglia (2014: 133) have pointed out, many observers, the ECB, and the European Commission agree that 'during crisis, clarity and speed in decision-making were crucial for bank crisis management' . However, the compromise solution of a multilayered SRM with many veto points is unlikely to deliver swift decision-making. Drawing on the successful crisis management experience in the 1990s, Swedish decision-makers favoured greater discretion for home supervisory authorities to move quickly and decisively when carrying out bank rescue and restructuring operations (Backstrom 1997;Englund 1999).
The bail-in mechanism introduced in the BRRD places the burden of bank losses onto shareholders and creditors before using any public funds. This principle aims to protect the interests of taxpayers vis-à-vis bank shareholders and mitigate moral hazard. It also provides guidelines regarding under what circumstances to conduct bank rescue operations and presents a more structured approach to resolving failing banks, while preventing panic and bank runs (Eliasson et al. 2014;Howarth and Quaglia 2014). Analysing the effects of the BRRD bail-in tool on the Swedish banking system, Eliasson et al. (2014) have raised concerns about potential contagion effects. In their view, indirect contagion effects, such as a sharp decline in market confidence, could be particularly severe in Sweden due to the closely interlinked banking system and Swedish banks' reliance on market funding (Eliasson et al. 2014: 43).
At the same time, under the BRRD, the Swedish national resolution authorities retain considerable discretion over bank restructuring and resolution. They can set minimum requirements governing how much capital and eligible debt instruments banks must hold on their balance sheets (Eliasson et al. 2014: 43). Furthermore, the actual resolution decision, including activating the resolution trigger, is still in the hands of the Swedish authorities (Eliasson et al. 2014: 36). By contrast, joining the SRM in addition to complying with the BRRD entails loss of regulatory discretion in crisis management. Under the SRM, the Single Resolution Board (SRB), the ECB, and the Commission take the leading role in managing cross-border bank resolution in the EBU (European Commission 2013a). The SRB can decide whether to use the Single Resolution Fund. The Council and national resolution authorities are consulted in the process, but their scope for intervention is very restricted.
In contrast to Sweden's decision to remain outside EBU, we argue that the Baltic states were not concerned about the decision-making rules or the recapitalisation regime. For all three countries, the overriding political decision was whether to join the eurozone or not. As Johnson (2008) has shown, the Baltic states were the among the pace-setters in the group of new EU member states in terms of fulfilling the Maastricht criteria to enter the European Monetary Union (EMU). Still, EMU conditionality, especially giving up domestic discretion in fiscal policy, was a contentious issue domestically (Epstein and Johnson 2010). In 2011, Estonia became the first Baltic country to enter the eurozone. In 2014, its central bank governor, Ardo Hansson, emphasised that the country's three-year eurozone experience 'affirmed the belief that the adoption of the euro was the right decision' (Hansson 2014). Furthermore, despite the severe economic contraction, neither Latvia nor Lithuania considered devaluation of their currency to be an option in order to meet the Maastricht exchange rate criterion (IMF 2010).
Once the ruling governments had established joining the eurozone as a key priority, the central banks and banking supervision authorities focused on explaining the benefits of EBU rather than seeking to opt out. Whereas some new EU member states from Central and Eastern Europe have been apprehensive about the dominance of foreign-owned banks, Baltic bank supervisors have been generally positive about the role of Scandinavian banks (Epstein 2013;Spendzharova 2014). The Estonian central bank, for example, has viewed the integration of the Estonian financial system with that of the Nordic EU member states as a strategic guarantee for financial sector stability (Eesti Pank 2009). At the annual conference of central bank governors of the three Baltic states, the governor of the Lithuanian central bank, Vitas Vasiliauskas, stressed that 'the single European currency is seen as a factor reducing the country's risk' (Cental Bank of Lithuania 2014a). In a subsequent statement, he also emphasised that by joining the eurozone, 'Lithuania is granted a voice in the ECB's highest governance structure' (Vasiliauskas 2014). For all three Baltic states, membership in the eurozone ensures having a voice in the SSM and the Supervisory Board of the ECB, which are ultimately responsible for the supervision of the largest eurozone banks.
Beyond the Nordic-Baltic region, how does the Swedish position relate to those of other non-eurozone countries such as the UK and Denmark? In the EU, UK decision-makers have been the most adamant about preserving national sovereignty and staying out of the EBU (Howarth and Quaglia 2013: 115). As Prime Minister David Cameron stated in a hearing at the House of Commons regarding the future of the eurozone, 'Clearly what is going to happen is if the eurozone is going to survive, it will have to make moves towards a more integrated state. We won't be part of that' (House of Commons 2012). By contrast, Swedish decision-makers have supported stronger crisis management capabilities at the EU level, but considered the uncertainty surrounding future financial liabilities to be too high to commit to joining the EBU from the outset (Backman and von Sydow 2015: 62; see also de Rynck 2015).
Just as in Sweden, Danish political parties have opposed paying for bank failures in other EU member states (Nielsen 2015: 16). Furthermore, Eurosceptic public opinion continues to constrain any deeper integration of Denmark in the EU despite the generally pro-EU stance of the Danish political elite. Thus, Danish decision-makers have taken a pragmatic 'wait and see' approach before they determine whether Denmark would join the EBU (Nielsen 2015: 22). Eurosceptic public opinion has played a role in the UK and Sweden as well. In a 2013 poll, British and Swedish citizens registered the highest level of disapproval when it comes to joining the eurozone (Backman and von Sydow 2015: 57).

'Banking union through the back door' for Swedish banks in the Baltics?
So far, we have seen that Swedish decision-makers opted not to join the EBU at the outset. Concerns about full participation in EBU decision-making and ability to preserve some national regulatory autonomy were important factors shaping this position. Nevertheless, the Swedish banking supervisory authorities will face constraints on their regulatory autonomy due to the tightly integrated Nordic-Baltic banking system. In practice, the systemically important divisions of Swedish banks in Latvia, Lithuania, and Estonia are subject to supervision by the Single Supervisory Mechanism, even though Swedish decision-makers chose to remain outside the EBU. Below we explain why implementing the EBU in the Baltic states significantly alters the scope of action of home supervisory authorities in Sweden, compared to the status quo before EBU. In our view, this leads to 'banking union through the back door' for Swedish-based banks in the Nordic-Baltic region.
In November 2014, the SSM became the new prudential supervisor of the largest eurozone banks, working in close cooperation with the national competent authorities (NCAs) of the EU member states. While the SSM supervises the parts of the banking group that are deemed systemically important, the NCAs are responsible for the remaining parts of these banking groups. For large cross-border European banks, this poses the challenge of aligning supervisory practices in cases when the host supervisor is in the eurozone but the home supervisor is not (Hansson 2014). This section unpacks the institutional and organisational implications of complying with EBU provisions for Swedish banks operating in the Baltic states and highlights the constraints on Swedish regulatory autonomy. In a regional perspective, 11 systemically important banks within the supervisory remit of the SSM are subsidiaries to five Nordic banks domiciled outside the eurozone (Central Bank of Lithuania 2014b). As of November 2014, the ECB has taken the role of lead supervisor of the subsidiaries of Swedish banks in Estonia, AS SEB Pank and Swedbank AS, and Latvia, AS SEB Banka and Swedbank AS (ECB 2014a). In Lithuania, the ECB took over direct supervision of the country's biggest banks in January 2015 when the country joined the eurozone and, consequently, the SSM. This affects the supervision of the country's three largest banks in terms of assets: SEB bankas, Swedbank, and DNB bankas (Central Bank of Lithuania 2014b).
Two coordination structures play an important role in institutionalising supervisory cooperation in the EBU -the existing supervisory colleges for cross-border banks introduced in the EU's Lamfalussy financial regulation framework (Quaglia 2010) and the newly set up joint supervisory teams (JSTs). Even before the introduction of EBU, the European Banking Authority was in charge of cross-border supervisory colleges as the permanent coordination structures for all national authorities involved in the supervision of a banking group (EBA 2014). These supervisory colleges will continue to play a role. Should, as in the case of Sweden, the home country authorities of the banking group be headquartered outside the eurozone, the ECB will also participate in the college. It will act as the host supervisor of the subsidiaries operating in the eurozone (Hakkarainen 2013;Nouy 2014).
By contrast, JSTs are new coordination mechanisms. They are the main supervisory body for systemically important banks in the eurozone. Each JST is in charge of the operational supervision of a single significant banking group within a eurozone member state (ECB 2014b: 9). Appointed for a period of three to five years, the teams are tailored according to the needs of the banking group, depending on its size and operations (Hakkarainen 2013). The chair as well as the core team (except sub-coordinators) should be ECB staff, and the full team is a combination of local and international staff from the ECB and NCAs.
Most importantly for our analysis, the ECB has extensive powers regarding the establishment and the composition of the JSTs. It can modify the appointments made by NCAs and has signalled its commitment to staff secondment and exchange in order to establish a common European supervisory culture (CEPS 2014). Furthermore, JSTs strive to achieve better balance between home and host countries in the supervision of cross-border banks. To safeguard against conflict of interest and bias, the JST chair cannot come from the same country as the home country of the bank. As Danielle Nouy, chair of the ECB Supervisory Board, pointed out, 'we will be a truly pan-European supervisor operating without national bias or prejudice' (Nouy 2014). She indicated that Crédit Agricole's chief supervisor will be a German national, Unicredit's will be from France and ABN AMRO's from Spain (Nouy 2014).
However, the division of eurozone insiders and outsiders has implications for the smooth functioning of supervisory colleges and JSTs. The Nordic-Baltic Banking Association, representing banks from Sweden, Denmark, Finland, Norway, Estonia, Latvia, and Lithuania, has expressed concerns about the actual operation of the JSTs. The association underscored the complexity of supervisory coordination within the teams and warned about unintended consequences of supervisory asymmetry between the authorities of member states inside and outside the eurozone (Nordic-Baltic Banking Association 2014). According to a task force report by the Centre for European Policy Studies (CEPS), JSTs will 'replace, at least for the SSM, the function of the Colleges of Supervisors of banks exclusively active in SSM countries' (CEPS 2014: 28). As highlighted by CEPS analysts, the ECB is facing a considerable operational challenge to acquire sufficient expertise about all 120 systemically important EU banks, compose balanced and competent supervisory teams, make the JSTs work effectively together, and avoid duplication of effort between the NCAs, ECB, and EBA (CEPS 2014: 28).
How will supervisory colleges and JSTs interact? The ECB has gained extensive supervisory powers with the introduction of EBU, but the EBA in London is not merely an information hub and standard-setter. It retains supervisory competences, especially for cross-border banks domiciled outside the eurozone. The EBA has stressed that it will continue to play an active role in banking supervision. It identified and monitored about 105 active supervisory colleges in 2013. Some 73 per cent of these colleges are headquartered inside the SSM, and the remainder outside the SSM (19 per cent) or in third countries (7 per cent). The EBA (2014) has emphasised that the number of supervisory colleges will decrease only slightly due to the SSM. Furthermore, only five banking groups will have a presence exclusively inside SSM countries, which requires a full transfer of supervisory responsibility to the ECB (EBA 2014). For the majority of large cross-border banking groups, the EBA will remain a relevant supervisory authority alongside the SSM.
The EBA Regulation was amended as a consequence of launching the SSM. Under the new model, the ECB will establish the supervisory college and serve as chair for the banks for which it acts as consolidating supervisor. This applies to the subsidiaries of Swedbank and SEB in all three Baltic countries. Bank supervisors from NCAs within the SSM can participate as observers. For significant banks outside the SSM, the ECB will participate in the supervisory college as a member and the NCAs as observers (Art. 10, ECB Regulation). The voting procedures in the EBA's Board of Supervisors were also modified to accommodate the demands of eurozone outsiders. The voting rules require qualified majorities among both eurozone insiders and outsiders in order to adopt EBA measures. Furthermore, the amended EBA Regulation allows a representative of the ECB's Supervisory Board to participate in the EBA Board of Supervisors (without a right to vote) and attend meetings of the Board of Supervisors pertaining to individual financial institutions (CEPS 2014: 37).

Conclusion
The Nordic-Baltic region including the EU member states examined in this article, Sweden, Estonia, Latvia, and Lithuania, provides a compelling case study of the reasons for the diverging choices of countries inside and outside the eurozone when it comes to joining the European banking union. We sought to explain Sweden's decision to opt out of the EBU. Inability to fully participate in the EBU decision-making process as a non-eurozone member state was a first concern for Swedish decision-makers. At the same time, unlike the UK, Sweden has not vocally opposed EBU. Secondly, Sweden was apprehensive about paying for failed banks in other member states. In a regional perspective, this was an important concern for Denmark too. Thirdly, the Swedish authorities preferred greater regulatory autonomy in crisis management to safeguard domestic financial stability. By contrast, for all three Baltic countries, the overriding political choice was whether to join the eurozone. As soon as their national governments took that decision, eurozone membership translated directly into EBU membership.
Still, we find evidence that the introduction of EBU does limit the regulatory autonomy of the Swedish authorities. The main constraint stems from the enhanced role of the ECB in the supervision of Swedish cross-border banks in the Nordic-Baltic region. We highlighted joint supervisory teams and cross-border supervisory colleges as central coordination mechanisms in banking supervision. The ECB has become the lead supervisor in the JSTs for systemically important Swedish subsidiaries in the Baltic states such as Swedbank and SEB. Furthermore, the ECB now participates as host supervisor in the supervisory colleges managed by the EBA even in the case of non-systemically important banks domiciled in countries outside the eurozone.