Beyond the North–South divide: transnational coalitions in EU reforms

ABSTRACT

The literature on fifteen years of European crises leaves the reader with a puzzle. Prominent accounts of the longest crisis – that of the euro area (EA) – assert that the EA is deeply divided between North and South, with Central Eastern European (CEE) member states being ignored. This makes it hard to explain how the union has managed to reform since 2008 and especially during the Covid-19 pandemic. Scholars have started to talk of transnational coalitions, but they equate coalition-formation with bringing together the like-minded, typically over solidarity versus sovereignty and more or less integration. However, coalitions of the like-minded are typically too small to sustain reforms and compromise has to be sought with others who have different preferences. To establish empirically how stable or fluid transnational coalitions are, we exploit the EMU|Choices database (Wasserfallen, Leuffen, Kudrna, and Degner 2019) [Analysing European Union decision-making during the Eurozone crisis with new data. European Union Politics, 20 (1), 3–23] on EA reforms and our own original data on Covid-19 reforms. Our findings show a stable pattern but no geopolitical divide – coalitions have varying CEE members. These findings can provide a basis for developing a more plausible conceptualisation of transnational coalitions.

The puzzle of institutional reform amidst political polarisation

One of the legacies of the euro area (EA) crisis is the notion that a deep divide between the North and the South of the union has taken hold, often with a normative connotation. Political economy scholars have provided ample evidence that deep divisions and structurally different interests marked the political debate between member states on what caused the crisis and the reforms necessary to end it (e.g. Johnston and Regan ²⁰¹⁶, Walter et al. ²⁰²⁰). But this raises the question how member state representatives from these two camps ever came to agreements that could keep the EA crisis at bay, even as other severe crises hit Europe, notably the Covid-19 pandemic. Scholars stopped at explaining why crisis is inevitable, not how it ended.

A European Banking Union was decided in 2012 and started in 2014; monetary policy became extremely loose and drove down risk spreads in Southern European bond markets. When the Covid-19 pandemic hit in early 2020, the stereotype of a North–South divide seemed to play itself out again but by the mid-2020s, the heads of state had agreed unanimously on a massive, redistributive recovery fund (Next Generation EU) that even earlier critics conceded was a surprise (Jones, Kelemen and Meunier 2021: 1526). How was this possible if the union is so hopelessly divided?

In line with a recent literature, we look for an answer in the role of transnational reform coalitions.¹ This literature is much more explicit on the policy preferences and package deals that underpin reform compromises than the ‘deep-divide’ school. But the two literatures share the assumption that governments as representatives of traditional country positions align themselves only with the like-minded. Buti and Fabbrini (2023) distinguish between the Frugal Four of the North, a Sovereignty coalition of Hungary and Poland as well as the Solidarity coalition of Continental Europe, but are not transparent on how they arrived at these classifications. Huhe et al. (2022, p. 79) are the most explicit. Using network analysis, the four authors build on previous work and assume an analogy of ‘homophily’ between actors and states regarding ‘’the tendency for people who share similar characteristics to form strong social ties’.

By contrast, we start from the common understanding of a coalition government in a domestic context (Warwick 1996). A coalition here is a temporary alliance of ideologically distinct parties for combined action. Before they make room for compromise, relatively homogenous groups of national representatives in support of a reform are likely to encounter resistance from other groups. To gain a majority, they typically must reach across the aisles. Hence, we expect transnational reform coalitions to include governments of member states with different, if not completely opposed, preferences.

The puzzle of institution-building amidst political polarisation arguably calls into question the standard, often implicit assumption of homophily in coalition formation. We examine this assumption and our counter-hypothesis in two path-breaking reform episodes over a decade. We are encouraged by Huhe et al. (2022) who do not find that ideological affinity has a robust influence, contrary to the approach in Matthijs and Blyth (2015). The presence or absence of populist parties in a government also makes no difference for forming ‘cooperation networks’, contrary to what the post-functionalist constraint predicted. Instead, we follow Wasserfallen et al. (2019) and start with positions on specific policy issues that can substantively and significantly explain the alliances between member state representatives. After all, policies are the focal points of democratic politics and we apply the same ‘policy-focused analysis’ (Hacker and Pierson 2014) to the EU’s orchestration of crisis measures.

More generally, we are interested in transnational coalitions as potential mechanisms of political participation and representation of national constituencies in the EU’s political system (Truchlewski et al. ²⁰²¹). Relatively stable transnational coalitions are Voice mechanism, analogous to parliaments in an intrastate context that the Rokkan-Hirschman model of polity-formation envisages (Rokkan et al. ¹⁹⁹⁹, pp. 228–33). Especially in crisis times, transnational coalitions can be a channel through which representation of national interests in reforms takes place and authority to decide over the reformed system is established. But we need to find out, firstly, whether the basis for such coalitions exist, namely compatible preferences regarding EU reforms, and then, secondly, how these coalitions encompass different preferences.

Our article proceeds as follows: in the next section, we review recent contributions on deep divisions and transnational coalitions in EU reforms. Then follows our empirical analysis of transnational reform coalitions during the EA crisis and the Covid-19 pandemic. We conclude by suggesting a parsimonious way in which one might make conceptual sense of these findings in future work.

EU divisions and transnational coalitions

In the political economy literature on European integration, it has become a stylised fact that the EU is deeply divided between Northern saintly creditors and Southern sinful debtors (Matthijs and McNamara ²⁰¹⁵). Even though scholars noting this division are critical of its normative connotations, they accept that this division explained why the currency union was in such a deep crisis, with the consequence of a hardening policy stance on both sides. The driving forces were Eurosceptic, typically right-wing, movements in the North that constrained German and Dutch governments in their support for EA reforms; in the South, left-wing anti-austerity movements threatened, and even drove out, governments in Greece, Italy and Spain which had to cut benefits and raise revenue, under market pressure and on demand from Northern peers. In accounts of comparative capitalism, this political polarisation was a symptom of heterogeneous economic structures that a common monetary policy cannot accommodate.²

This original Varieties of Capitalism diagnosis is hard to square with the fact that Ireland was afflicted by the same vulnerability to a financial crisis as most of Southern Europe, while Italy was not. More refined growth models take on board that Southern European growth was fuelled not by wage inflation but by credit to middle classes aspiring home ownership. Blyth (2015: ch.3) provided an ideational power diagnosis that is compatible with the growth models tradition: an ordoliberal obsession with stability makes Germany and EU institutions responsible for imposing a self-defeating austerity regime on the EA. On the back of rising inequality, financial markets came to the rescue by handing out credit lavishly. The boom–bust cycle that ensued was the paradoxical result of this drive for stability in an institutionally incompatible union of members. Baccaro and Tober (2021) specifically argue that wage restraint of Southern European labour could not have compensated for the inflationary consequences of high capital inflows, while less wage restraint in Germany would have reduced its export surplus. Frieden and Walter (2017) can explain why this constellation was so stable, by disaggregating ‘country interests’. They argue that the credit boom was popular and stopped only with a balance-of-payments crisis of the South. Their approach makes them see complementarity of dominant sectoral interests across member states, for instance export industries in the North and non-tradable services in the South.

The growth models line of crisis diagnosis would lead us to expect that member states coalesce on reforms with those that are on the same side of the materially and ideationally determined divide. But they do not analyse how coalescing happens. Hegemony can be a way of bringing in the political role of transnational coalitions in EU reforms and crisis management. The contributors to Schoeller and Falkner (2022) study coalition formation of small states as a response to an unpredictable German hegemon. The New Hanseatic League (NHL), an alliance of eight member states in Northern and Eastern Europe, is the prime example (Schoeller 2022). More generally, Schoeller’s hegemonic instability theory explains country coalitions as a temporary phenomenon that unleashes centrifugal dynamics. The NHL proved to be fragile indeed, playing no role in recent Covid-related reforms. These reforms saw the alleged hegemon, Germany, to change its position even more in the direction resented by NHL members. This should have mobilised their alliance. Moreover, in five of the seven country cases contained in Schoeller and Falkner (2022), governments were ‘sitting on the fence’ instead of opposing Franco-German dominance: only Austria and the Netherlands were vocal, supposedly close allies of Germany. Arregui and Thomson (2009) provide evidence that smaller member states tend to focus their engagement on issues that matter to them, which ensures their relatively high success rate. This does not bode well for a hegemonic theory of transnational coalitions.

Political scientists analyse the revealed preferences of national representatives in EU reforms (Fabbrini 2023, Schoeller and Falkner ²⁰²²). In analogy to the theory of domestic cleavages, these scholars formulate values that the sides of an underlying conflict share and express in public. In the context of European integration, it is common to identify the preference for less integration with the quest for sovereignty and the preference for more integration with the quest for solidarity. Fabbrini (2023) conceptualises these preferences as the extremes of a spectrum. The Frugal Four, at one end, emphasise national fiscal responsibility and solidarity that is confined to specified instances of insurance beyond a member state’s control. This is different from the ‘sovereignty coalition’ of Hungary and Poland that share the national responsibility in every respect but do not oppose solidarity in the guise of means-tested transfer programmes. They are in favour of less and more integration, depending on the policy issue.

We take analytically another route. Huhe et al. (2022, pp. 82–3, 89-90) find that ideological similarity on EU integration and the market economy is only a weak predictor for stable cooperation in networks of national representatives. Positions on specific policy issues make the same countries cooperate repeatedly. Hence, we classify specific policy reforms for the issues of sovereignty or solidarity that they raise.³ This route can arguably explain affinities between countries rather than deterministic divisions and can also make sense of the discernible stance that CEE countries take (de la Porte and Jensen ²⁰²¹, Fabbrini ²⁰²³).

Our conceptualisation of solidarity is broader than implied by demands for a transfer union that confine solidarity to grants by the rich for the poor (Genschel et al. ²⁰²¹). This excludes insurance of the unfortunate by the fortunate, which we include. Reforms are about (more) solidarity if they create and extend the risk pool of EA member states, increasing interdependence. Reforms are about (less) sovereignty if they affect states’ rights to self-determination vis-à-vis EU institutions, i.e. they must not decide unilaterally on policies or non-action, for instance because of the externalities for other members.

We analyse reforms that stretch over nearly fifteen years, including the Covid-19 reforms. In a situation of intense problem pressure, the urgency of decision-making privileges tried-and-tested alliances in which members share an understanding of sovereignty and/or solidarity. But they may share an idea of solidarity on a reform that is mainly concerned with mutual support, yet not share an understanding of sovereignty when a reform is about restraining national discretion. This is one way of explaining why governments shift alliances, albeit not randomly, a conjecture to which we come back in the conclusions.

Empirical analysis of transnational coalitions in EU reforms

Our analysis proceeds in three steps, triangulating different data sources. First, we look at country positions, i.e. revealed preferences, concerning reforms of economic governance that took place at the height of the EA crisis in 2010–2015. We adapt the coding scheme of Wasserfallen et al. (2019) of more or less integrative reforms to reforms concerning solidarity or sovereignty, respectively. Second, we reproduce this exercise of finding potential coalitions for the Covid-19 reforms, mindful of the fact that they happened over a shorter time span. This should privilege alliances between the like-minded if such alliances are ever prevalent. Third, we explore whether publicly known coalitions in the two reform periods – for instance the Frugal Four and the NHL – corresponded closely to similar revealed reform preferences in these two crises. In all these exercises, we make methodological choices that favour the claims in the political economy literature, notably finding a stable North–South cleavage with CEE member states as by-standers. If we do not find dichotomous divisions and repeated stand-offs between the usual suspects, then we have reason to believe that these diagnoses need qualification.

Classification of reforms in terms of solidarity and sovereignty

We classified reforms in the following way: Solidarity reforms require states to underwrite more reciprocal risks (i.e. higher levels and/or a larger set of risks). Intensified interdependence materialises as contingent liabilities for each state’s budget: for instance, if the size of a bailout fund increases, or if not only the risk of a bond market panic is covered but also the evaporation of capital in a national banking system. More controversially, reforms that institutionalise the risk pooling with third parties is also a Solidarity reform, notably the involvement of the IMF as an organisation and all its underwriting members, or of private shareholders of banks through a bail-in clause. So, in the case of a risk that materialises for some unfortunate members of the EA, a fortunate member state has to show solidarity as a fiscal authority, a member of the IMF, and as a jurisdiction of bank investors.

Sovereignty reform affect national discretion in deciding a policy, for instance when the Commission implements the European Semester or the ECB can influence the restructuring of banks. Sovereignty is also affected if governments must comply with recommendations of national fiscal councils that are mandated by EU legislation or intergovernmental agreements. Here, national executives commit to sharing their sovereignty in the sense that they must not decide unilaterally on policies or on non-action when this can have externalities for other members. Note that this does not imply that governments are always opposed to this. In a diverse union with power asymmetries, restraining national sovereignty of others can also imply relative empowerment, notably for smaller member states or those with a weak bargaining position.

There can be a Trade-off between solidarity and sovereignty if a reform links the extension of risk-sharing with a constraint on the autonomy over domestic affairs or grants more national discretion only in return for limited solidarity. Conditionality is a case in point: bailout funding can be made conditional on institutional reforms and severe cuts in fiscal spending; in turn, the conditionality for a bank restructuring programme comes with less intrusive conditionality but the funding is also less fungible for the receiving government. The trade-off between sovereignty and solidarity is often couched in terms of risk reduction/prevention versus risk sharing. We expected this to be quite a powerful narrative and framing of reform issues.

Reforms during the EA crisis

The preferred method for studying coalitions between member states used to evaluate their voting behaviour in the Council (Naurin and Lindahl 2008, pp. 65–7). But this underestimates the extent of coalition-formation as many decisions never come to a formal vote in the Council. Moreover, holding a similar initial position to another member does not mean that the two governments are in a coalition just as holding different positions does not preclude forming a coalition. Hence, we follow the approach of Naurin and Lindahl (2008), who study coalitions according to the revealed preferences of governments throughout the process of decision-making. This requires datasets based on expert interviews assessing the original expressed position of governments and their support for a reform later which they coordinated with others.

Our search for country coalitions around a sovereignty-solidarity conflict line starts with the EMU Positions data (Wasserfallen et al. ²⁰¹⁹).⁴ These scholars sub-divided ten reform packages into 47 contested reform issues negotiated during 2010–2015. Wasserfallen et al. (2019) assign a value of 0 to reforms that imply no further integration beyond the status quo and 100 for the most integrationist position debated, possibly with one or two positions in between. By classifying EMU|Choices reforms in terms of solidarity and sovereignty, we distinguish reforms in the binary way that the literature classifies coalitions: if they are right, Northern members (the ‘sovereignty coalition’) should largely oppose solidaristic reforms and Southern members (the ‘solidarity coalition’) should largely oppose reforms that strengthen sovereignty. Table A1 in the Appendix summarises our classifications.⁵

First, of 47 reforms, 44 can be classified as concerned with either solidarity or sovereignty. Seven of eight reform packages were primarily about solidarity or sovereignty; the European Stability Mechanism (ESM) could not be categorised: three of the seven issues were about a trade-off of sovereignty, four about solidarity. The first bailout programme for Greece and the EFSF (European Financial Stability Facility, predecessor to the ESM) were about (limits to) solidarity. Five packages were predominantly about more or less sovereignty: the Six-Pack and Two-Pack, the Fiscal Compact, the Banking Union and the Five Presidents’ Report. They limited (further) national discretion over fiscal policy and financial supervision. The reform packages concerning sovereignty tended to come later.

No reform package was entirely about the trade-off. Rather, reforms negotiated simultaneously could be interpreted as presenting such a political exchange. For instance, the first Greek bailout and the EFSF concerned solidarity but were accompanied by the Six-Pack reforms concerning sovereignty. Surprisingly, only three issues were framed as a trade-off between solidarity and sovereignty. This was twice the case in the permanent bailout reform package of the ESM (access to financial assistance only against tighter surveillance) and once in the Six-Pack (withholding EU funds to countries with excessive budget deficits). A direct trade-off is rare and was agreed only once: access to the ESM has been predicated on signing the Fiscal Compact. While these are significant cases, it suggests to us that, contrary to our expectations, a direct trade-off is not an accepted framing of reforms.

Country clusters

By identifying country clusters, we want to find out whether there were stable preference clusters with a clear North–South divide at the opposite ends. The original EMU|Choices dataset has five countries (Croatia, Bulgaria, Hungary, Lithuania and Romania) with at least 20 per cent of missing values and some reforms have less than half of all members positioning themselves. It is a methodological choice to treat this as non-information and exclude such cases or to treat them as information. Especially governments of smaller members may be less invested in the matter and feel well-represented (Arregui and Thomson 2009).

Figure 1 visualises our clusters with the full set, taking absence of a revealed position as information still. From left to right, the dendogram shows the distance between countries: the closer the countries are on particular reforms, the shorter its legs. Each bracket shows a grouping with countrie­s that are decreasingly similar as the brackets move outward. We recoded the values of the EMU|Choices database (more or less integration) such that darker shades mark more solidaristic positions in Solidarity reforms and less insistence on national responsibility in Sovereignty reforms, respectively.

Figure 1.

Cluster analysis of EA reforms. Source: EMU|Choices database recoded.

Note: Appendix A1 contains more details on our coding. The colour coding for the top row below the countries shows the monetary status of each country: dark grey for early adopters of the Euro in 1999 (e.g. Spain or Austria), light grey for late adopters of the Euro (e.g. Lithuania in 2015) and white for countries outside the Euro Area (e.g. Czechia). The colour coding on the rightmost column next to the reform abbreviations reflects their classification as solidarity (light grey) or sovereignty (black) reforms, and trade-off (dark grey). The colour coding for the main cells follows a grey scale from 0 (light grey) to 100 (dblack) for each solidarity and sovereignty reform. A white cell shows missing information.

Figure 1 shows that two distinct clusters, which at first sight seems to support the ‘deep-divide’ literature. The first cluster of nine countries from Italy to Cyprus comprises a third of the sample. It contains the big EA members except Germany (France, Italy, Spain), and all EA countries that needed an ESM programme, including Ireland. The latter’s inclusion as well as the presence of Belgium and Luxembourg defy the stereotype of ‘Southern-Latin’ preferences. We label it the solidaristic cluster because members tend to support consistently any form of more risk-sharing (solidarity).

The second cluster comprises eighteen countries divided into three sub-clusters. Twice as big as the solidaristic cluster, it is also less coherent. The sub-cluster next to the solidaristic cluster includes Estonia, Slovakia, Slovenia, and Germany. This cluster is not an artefact of the method since these four countries are among the most vocal position-takers, joined by Austria, the Netherlands and Finland. A second sub-cluster in the middle consists of eight Nordic, Baltic and Eastern countries, plus Malta. Several are Euro-outs: presumably, these countries do not wish the EA to share ever more risks, possibly roping them into such interdependence. The polar opposite to the solidaristic cluster consists of a similar geographic mix, and includes two original EA members, Finland and the Netherlands.

It is possible to read a territorial/geopolitical dimension into our findings, however, not between North and South but East and South. All Southern states are in the solidaristic cluster, except Malta. It includes no CEE member, noteworthy for the fact that Latvia, Hungary and Romania received bailout programmes before Greece. There are also important outliers. One of them is Germany, surprisingly apart from where the literature would place it (to the right). Ireland is in the solidaristic group but only marginally as different robustness checks of the full set cluster analysis show (Table A2 in the appendix). The combination of ideology and interests makes Eastern governments (and Malta and Ireland) flexible and unpredictable.

Reform clusters

Figure 1 also shows polarising reforms, i.e. reforms on which the members in each identified country cluster are close and the coherent position of each cluster is quite distinct from those of others.

Five out of the six polarising issues we identified concern solidarity. The only exception is BU7 which raised the sovereignty question of decision-making powers in the Single Resolution Mechanism of the Banking Union: the solidaristic cluster was consistently in favour of having the resolution authority centralised in the Commission, while other sub-clusters were in favour of an intergovernmental body controlled by member states. The final agreement, not part of the initial negotiations, limits the involvement of member states to disagreements between the Single Resolution Board and the Commission, tilting it towards the position favoured by the solidaristic cluster.

Reforms to provide bailout funding were extremely polarising. Concerning G1 (the initial willingness to support Greece), coherently in favour were the solidaristic cluster and the mostly non-EA countries in the cluster around Lithuania and Hungary⁶; opposed were the clusters around Germany (except Slovenia) and the sovereigntist group, except Latvia that had a programme before. ESM5 concerned the question whether the bailout fund should be given more and wider instruments beyond loans to sovereigns. The solidaristic cluster was largely in favour of allowing the ESM to buy bonds directly from sovereigns and finance direct bank recapitalisation (although Ireland and Cyprus disagreed) while the cluster including Germany was opposed to both widening and flexibility. The two sub-groups around Lithuania-Hungary and the sovereigntist cluster, respectively, have more missing data than revealed preferences. The agreement in 2011 favoured the position of Germany et al.

Elements of a regular fiscal union were also polarising. TPA1 concerned a redemption fund with joint liability that a majority in the European Parliament proposed. The solidaristic cluster was in favour (missing data for Belgium and Cyprus), two sub-clusters are consistently opposed, while the sub-cluster around Lithuania and Hungary shows the whole range of responses: support, opposition, and silence. In other words, it was not a coalition on this reform issue. Interestingly, there was no such polarisation as regards EB1, the proposed introduction of Eurobonds. Both proposals, TPA1 and EB1, remained undecided. A similar issue came up in the context of the Banking Union (BU9), namely whether the ESM should be allowed to back the Single Resolution Fund (financed by banks themselves). The solidaristic cluster was in favour (data for Luxembourg is missing), the cluster around Germany and the sovereigntist preference group wanted to give the Fund only the permission to borrow while the cluster around Lithuania and Hungary is again internally divided and largely silent.⁷ Finally, support for a Financial Transactions Tax (FTT) found near-consistent support from the solidaristic cluster (three countries had other preferences) and from the cluster including Germany; there was consistent opposition from the other two sub-clusters, with the exception of Austria. It is noticeable that there is no missing data on the FTT. The proposal for such a tax passed under ‘enhanced cooperation’ with an avantgarde of countries but was never implemented (Kalaitzake 2017).

COVID-19 pandemic reforms

We followed the process of EMU|Choices scholars to code the reforms undertaken during the pandemic 2020–2021. First, we systematically collected and coded newspaper reports using policy process analysis (Bojar et al. ²⁰²³), a novel method that codes the entire process of a policy proposal, the prior position of countries on a reform, the debate and the outcome. Second, we systematically collected and analysed official documents to gather additional information and adjust our inferences of country positions. Finally, in the spring and summer of 2021, we interviewed European Union officials close to the negotiations, who could confirm or correct our coding of country positions (see list in Appendix A3).

We thus coded 13 reform proposals in four reform packages:

• the quest for debt mutualisation in a letter signed by nine governments (one reform: CBND_OFF for Coronabonds official letter of the 9 countries);

• the reform of the ESM after a hostile reaction in Italy and Spain to loans with conditionality (three reforms: ESM-PCS_1 on the use of the ESM or not; ESM-PCS_2 on the nature of conditionality; and ESM-PCS_3 on who would be eligible for the credit line);

• the Recovery and Resilience Facility (eight reforms: RRF_1 using grants and/or loans; RRF_2 conditionality and reform of fiscal rules as a pre-condition, RRF_3 allocation to sectors/regions, authorities with plans or member states with needs; RRF_4 strictness of the rule of law conditionality; RRF_5 financing of RRF through guarantees for national bond issues, Commission own resources or perpetual bonds; RRF_6 liability of each member state, secured by EU budget, or joint; RRF_7 size of €250 billion, €500 billion or €1.5 trillion⁸; RRF_8 eligibility according to severity of shock and/or low income);

• and finally the European instrument for ‘temporary Support to mitigate Unemployment Risks in an Emergency’ (SURE, one reform).

We consider eight reforms, as being about Solidarity, risk-sharing for the employment shock of lockdowns (SURE) and for an uncertain recovery (Coronabonds and the RRF). Five reform issues are classified as concerning Sovereignty: the reforms of the RRF_2,3,4 and ESM-PCS_1 and 2 (see above).

Given how intense the reform period in 2020 has been, it is difficult to reproduce the EMU|Choices dataset in its detail of information, where reforms were discussed as packages over longer periods, providing incentives to state one’s preferences. The pandemic forced EU and member state officials to negotiate online, so few official documents are available for coding the initial position of countries. Consequently, our database has a high level of missing values, 45% compared to the EMU|Choices dataset with between 25 and 30% of missing information. This concerns especially the ESM and SURE reforms; Estonia was completely silent. Our interviewees suggested that some countries did not bother to state an official position, either because they were too busy with the pandemic at home or because they were not particularly concerned.

Our cluster analysis of the whole dataset takes all reforms into account (Figure 2) and we note again two clusters. However, the left-hand, ‘solidaristic’ preference cluster is extremely heterogeneous, while the right-hand side cluster assembles the Nordic, Baltic and Eastern group of smaller member states. The German government took a middle-of-the-road position on almost every issue.

Figure 2.

Cluster analysis of four Covid-19 reforms. Source: own data.

Note: Colour coding same as in Figure 1; full details in Appendix A3.

The preference clusters become somewhat clearer if we take only the RRF reform package which we classified as one of solidarity. Our data is relatively complete. This gives us three clusters: a still heterogeneous solidaristic group of fourteen countries; a homogenous group of the Visegrád-4 plus Slovenia, that is divided from the rest over RRF-4, the rule of law stipulation; another semi-homogenous group of seven countries comprises the Nordic fringe including Lithuania, as well as Austria, Germany and the Netherlands. As noted, however, Germany has a fairly middling position that can go either way of the spectrum: it switches to the cluster on the right when we look only at RRF reforms (Figure A1 in the Appendix).

Our conclusion is that revealed reform preferences do not cluster very closely and coalitions synthesise different preferences. Again, however, no Southern European country is in the cluster on the right and no CEE member state is in the cluster on the left, a pattern that we have already seen with respect to EA crisis-related reforms.

From preferences to reform coalitions

How closely do the revealed reform preferences we identified correspond to publicly known reform coalitions? While not claiming that the list is complete, Table 1 lists five known country groupings that have taken a public stance on certain EU/EA reforms.

Table 1.

Transnational coalitions and the reform preferences of their members.

Revealed preferences Reform coalition	EA reforms 2010–2015	Covid-19 reforms 2020
Franco-German couple (FRA, DEU)	Different preferences on almost every reform (except SPA1; FC1, 8, 9; PR1, 3, 4; TFF)	Same preferences on almost every reform (except CBND-OFF)
New Hanseatic League (CZE, DNK, EST, FIN, IRL, LVA, LTU, NLD, SVK, SWE)	Different preferences, in three of four sub-clusters	Different preferences, in either of the two clusters (IRL, LTU, LVA in ‘solidaristic’ cluster)
Frugal Four (AUT/ FIN, DNK, NLD, SWE)	Somewhat different preferences, in two of four sub-clusters	Same preferences, in most ‘sovereigntist’ (sub-)cluster
Visegrád Four (CZE, HUN, POL, SVK)	Different preferences, in three of four sub-clusters	Same preferences, next to ‘sovereigntist’ sub-cluster
Coronabonds-9 (BEL, ESP, FRA, GRC, IRL, ITA, LUX, PRT, SVN)	Same preferences (missing information for SVN)	Same preferences (missing information for SVN)

Source: cluster analyses of Figures 1 and 2.

There is, first, the Franco-German couple which plays a crucial role in EU reform processes, documented by the literature on hegemony and leadership in European integration (Schoeller 2022, Krotz and Schramm ²⁰²²). The two typically started from opposite positions and tried to come up with a compromise before decisive summits. Over the long decade of crises management and institutional integration, there have been three French Presidents (Sarkozy, Hollande and Macron) that worked in very different ways with the German Chancellor Merkel. While Sarkozy came round to support a heavy-handed agenda-setting role of the two countries, e.g. with the infamous Deauville Declaration, Hollande’s administration adopted openly the role of a counterweight to Germany’s ideological and institutional dominance. Macron has used his close personal rapport to influence Germany’s position in a more integrationist-solidaristic direction. Macron’s success can be inferred from the preference convergence in 2020, when reforms took a decisive turn towards more solidarity. The Coronabonds proposal is a noteworthy exception, which had to be abandoned by the French President.

The NHL since 2018 was responsible for the interest in transnational coalitions in EU reforms in the first place (Schoeller 2022, Verdun ²⁰²²). In March 2018, Denmark, Estonia, Finland, Ireland, Latvia, Lithuania, the Netherlands, and Sweden published a letter challenging the Franco-German initiative of EU reform, raising eyebrows in Brussels and Paris (Khan 2018). Joined by Slovakia and Czechia, they insisted that the Single Market is the core of European integration. Hence, their nickname. They were prominent in blocking in particular Macron’s attempts at creating additional fiscal capacity for the EA in 2019. The coalition was, however, of rather short-lived existence as Dutch and Scandinavian governments supported the rule of law stipulation that the Visegrád Four saw as undue interference. But even before, it was not a coalition of the like-minded. On EA reforms (Figure 1), the NHL members were in three different sub-clusters other than the solidaristic cluster. Our analysis suggests that we should not be surprised by its short-lived existence.

The Frugal Four or Five are a relatively old formation, originally with Finland, but during the pandemic its social-democratic government was replaced by Austria (Vatanen 2020). It is an initiative of Dutch governments, trying to act as a counterweight to the French influence on Germany (Verdun 2022). While their preferences have not always been completely aligned on EA reforms, their preferences have converged during the pandemic. This development makes sense against the backdrop of Germany accommodating France’s position, which was on the extreme of the solidaristic cluster.

The Visegrád Four has been an alliance since their summit in 1991. During the EA crisis, it became a voice for ‘a Europe of Nations’ (Ágh 2018) that resented ever closer (monetary) union. They were members of the NHL, but their populist-authoritarian and traditionalist turn has estranged the rest of the EU from them, making them more coherent in turn. The latest stipulation of linking recovery funding to compliance with the rule of law (RRF_4) united them (and Slovenia) further against what they see as excessive Western liberalism and cosmopolitanism. Even so, in terms of revealed preferences they are not in the most sovereigntist preference sub-cluster.

The Coronabonds-9 was given this name after they sent a letter to Council President Michel in March 2020 that asked for debt mutualisation. It is a large group and ever since 2009–2010 remarkably consistent in their quest for more mutual support in the EU/ EA. But until the pandemic, this country grouping of euro-insiders had not appeared as a political voice channel. It is also remarkable that France was a signatory of the letter, only to propose in May 2020 together with Germany a temporary EU recovery fund of €500 billion grants (Krotz and Schramm 2022, Schelkle ²⁰²¹).

A noticeable feature exposed by Table 1 is that preference clusters have become more overlapping with reform coalitions: the Franco-German couple, the Frugal Four, and the Visegrád Four had similar preferences in 2020, but not in 2010–2015. Obviously, the EA reforms concerned 47 and not just 13 reforms, that were less important to euro-outsiders, and the packages consisted also of more reforms that affected sovereignty. But it suggests that the policy response to Covid-19 was contested, contrary to the widespread impression that the nature of the pandemic favoured consensus (e.g. Ferrara and Kriesi ²⁰²²).

Our findings suggest that macro-regional short-cuts in terms of Southern, Northern or Eastern preferences hide how internally differentiated clusters are: Belgium and Luxembourg express regularly, Ireland sometimes, ‘Southern preferences’; Baltic and most Eastern European countries often share the positions of Scandinavian countries, which the Netherlands consistently takes as well. Governments from Germany, Ireland, Malta and Slovenia are unpredictable if reform choices are framed along one-dimensional preferences for sovereignty or solidarity. Hence, using geopolitical terms for transnational coalitions can be misleading when explaining how the EU reformed against the odds of political division.

Discussion and conclusion

Explaining massive institution building during severe crises preoccupies an army of scholars studying the EU polity. Our paper participates in this struggle. Comprehensive reforms proceeded despite real and stable divisions. How where these divisions overcome to arrive at an agreement, typically a compromise? Macro-regional divisions are too deterministic. ‘North’ and ‘South’ do not capture uniform and united political economies. Even if they formed a camp each, Central Eastern Europeans cannot be subsumed under either, yet they matter since they can shift majorities. One clear divide is that Southern European countries are almost always on the solidaristic side and in favour of more EU competences while CEE member states have the exact opposite preferences. ‘The North’ is made up of countries with relatively stable but varied preferences, and Germany is a surprisingly unpredictable outlier.

How can we make general and not only crisis-specific sense of our empirical findings? We see two avenues for future research. One way of making sense of the varied – but not randomly changing – reform coalitions could be to consider domestic constraints. Researchers could draw analogies to the economically motivated constraints of sectoral interests (Frieden and Walter 2017) and trace the political-ideological conflict that the crisis management in the EU regularly activates. EU reforms force governments to argue for more or less interdependence through institution-building, which is a tall order in the midst of a crisis. This made scholars formulate a transnational cleavage of demarcation and integration in domestic politics (Kriesi 1998, Kriesi et al. ²⁰⁰⁶, Hooghe and Marks ²⁰⁰⁹), in addition to the left-right cleavage. Hooghe and Marks (2018) suggest that the new cleavage created a ‘constraining dissensus’ on European integration, reversing the ‘permissive consensus’ of the early post-war decades.

But depending on the perceived incidence of a crisis, even hard-core Eurosceptic parties have demanded more EU support in vulnerable Italy and mainstream pro-European parties favoured national responsibility in the Netherlands. The transnational cleavage has been repeatedly superseded by how countries were affected in a specific crisis. Moreover, the urgency of stopping an escalating crisis eventually calls for collective action and at that point, most voters want to see the escalation stopped or the crisis contained (Ganderson et al. ²⁰²³). This can explain why the network analysis of Huhe et al. (2022, pp. 92–3) found no evidence for the presence of populist parties having a predictable effect on cooperation patterns; they conclude ‘that EU-level cooperation networks are to a certain extent insulated from national politics’. Democratic responsiveness and representation still requires governments to synthesise and select which voter concerns they express at the EU level (Mair 2009). We see the ongoing reform process in the EU, despite articulated diversity of preferences, as evidence corroborating a modicum of strategic autonomy rather than a one-sided constraint on governments.

An alternative to a pure electoral constraints explanation is to allow for different understandings of solidarity and sovereignty among elites. It amounts to a minimal and plausible qualification of the existing literature on transnational coalitions: we should assume that even governments that are firmly in favour of solidaristic reforms value sovereignty; and vice versa, steadfast sovereigntist nations may not resent mutual support under any circumstances. The signatories of the Coronabonds letter resented a repeat of ESM bailouts for treating countries as semi-sovereign; and it was Dutch PM Rutte, who first proposed a small charitable fund for member states hardest hit by the pandemic (Schelkle 2021).

Sovereignty can mean collective empowerment of members by a community and it can mean national self-determination; the literature takes the latter as the only meaning. Analogously, solidarity can mean to share risks or to share resources, that is, be insurance for bad luck from which even the well-off member states may benefit or be redistribution to those with predictable disadvantage like poverty or arrested development, that is bad luck suffered in the past. Again, solidarity as redistribution is the meaning that the diagnosis of a North–South divide presupposes. In our cluster, the consistent stance of Belgium and Luxembourg siding with the solidaristic camp is consistent with a notion of sovereignty as collective empowerment and solidarity as sharing risks. A parsimonious conceptualisation of at least two different ideas of solidarity and sovereignty allows for divisions but can also explain more varied outcomes: those who share an understanding of sovereignty and coalesce on those issues may still disagree when it comes to solidaristic reforms. These varied and compatible but different understandings can result in innovative institution-building despite a constraining dissensus, solving the puzzle of EU polity formation in hard times.

Funding Statement

This research was supported by the European Research Council under the Synergy Grant number ERC_SYG_2018 grant number 810356, in the scope of the project SOLID – Policy Crisis and Crisis Politics. Sovereignty, Solidarity and Identity in the EU post 2008.

Disclosure statement

No potential conflict of interest was reported by the author(s).