Differentiated integration in the European Union: Institutional effects,public opinion,and alternative flexibility arrangements

Abstract

Research on differentiated integration (DI) in the European Union (EU) has focused on the causes, conditions, and patterns of differentiation in European integration. By contrast, we know less about its effects on institutional outcomes and public support; moreover, alternatives to de jure DI in providing flexibility are still rarely accounted for. This introduction to the special issue takes stock of, and discusses omissions, in the current literature on DI. We propose an analytical framework, centering on efficiency and legitimacy, to study the effects of different types of DI. We use this framework to motivate the choice and assess the contributions of the articles selected for this special issue.

Introduction

Research is quickly catching up with the reality of differentiated integration (DI) in the European Union (EU). Transitional arrangements granting member states temporary derogations from the application of European legislation have been in use since the very beginnings of integration into the European Economic Community. Such transitory differentiation has enjoyed particularly frequent use in the context of the accession of new member states. Since the 1990s, the EU has further experienced a series of open-ended differentiations for extant member states. The best-known examples are the Eurozone and the Schengen Area: both have exempted members unwilling to participate, and both continue to exclude members that do not meet the prerequisites. In parallel to this internal differentiation, the EU has created a variety of external differentiation schemes, which allow non-members to participate selectively in EU policies, starting with trade and association agreements in the 1960s and 1970s and culminating in the European Economic Area (EEA) that entered into force in 1994.

Academic work has accompanied the growth of DI, but it initially remained predominantly conceptual and prescriptive, suggesting and categorizing alternative ways in which the EU could develop away from the model of uniform integration (see Holzinger and Schimmelfennig, 2012 for an overview). By contrast, the major general theories of European integration did not pay systematic attention to differentiation, and theory-guided empirical work was extremely rare—Kölliker's (2001, 2006) application of collective goods theory to DI being the main exception.

In the past decade, the situation has changed massively. Recent research has included differentiation in the theory, definition, and measurement of integration and has engaged in systematic data collection and analysis (Duttle, 2016; Duttle et al., 2017; Leuffen et al., 2013; Schimmelfennig and Winzen, 2014, 2019, 2020). Related works have explored other forms of flexibility in EU legislation and implementation beyond the legal exemption or exclusion of member states (Franchino, 2006; Steunenberg and Toshkov, 2009; Zhelyazkova and Thomann, 2022).

DI research has received a further boost from contemporary developments in European integration as well as from EU funding support. Whereas the reform debate on the Future of Europe included internal differentiation as one option (Commission, 2017), the Brexit negotiations have put external differentiation in the spotlight. Moreover, the EU's Horizon 2020 program for research and innovation has funded four collaborative projects on DI to advance the research agenda and consolidate research networks. ¹ In short, DI has shifted from a marginal to a mainstream topic in EU research (Leruth et al., 2019, 2022).

At the same time, systematic blind spots remain. Research has focused on the causes, conditions, and patterns of differentiation in European integration. It has sought to explain why member states choose differentiation, under which conditions DI is more or less likely to occur and endure, and how it develops over time and across countries (Schimmelfennig and Winzen, 2020; Schimmelfennig et al., 2015).

By contrast, we know less about its institutional and public opinion effects. Does DI facilitate EU decision making, strengthen democratic procedures, and improve the implementation of EU law? Do citizens approve of DI, and does differentiation improve public support for European integration more generally? Existing research often starts from the implicit or explicit assumption of rational actors choosing DI in anticipation of its legitimacy and efficiency benefits (Kölliker, 2006; Schimmelfennig and Winzen, 2020). Moreover, political theories of DI are based on normative assumptions about legitimacy and efficiency (Bellamy and Kröger, 2021; Lord, 2015). It is not clear, however, whether these assumptions can be empirically substantiated. In this special issue, we therefore probe into the institutional and attitudinal effects of DI. Moreover, we compare DI to alternative forms of flexibility in integration.

In the remainder of this introductory article, we present core concepts, empirical patterns, and explanatory factors of DI and then discuss omissions in the current literature, most notably relating to the effects of DI. We propose an analytical framework for the study of effects of DI centered on efficiency and legitimacy. Finally, we use this framework to discuss the findings of the special issue articles.

DI: Concepts, patterns, and explanations

Types and modes of DI

DI (as opposed to uniform integration) refers to a situation in which the territorial extension of the legal validity of EU rules is incongruent with EU membership (Holzinger and Schimmelfennig, 2012: 292). Put differently, European integration is uniform if all member states, and member states only, formally participate in EU policies (understood as the set of EU rules pertaining to the same issue). If at least one EU member state does not participate in an integrated policy, we speak of “internal differentiation,” and if at least one nonmember state “opts in”, we speak of “external differentiation” (Leuffen et al., 2013).

To complicate things further, in addition to this “horizontal differentiation”, the EU also displays “vertical differentiation”–as in federal states, EU policies are differentially centralized, ranging from exclusive EU competencies exercised by supranational bodies (as in the monetary union) to exclusively national competencies that are, if at all, intergovernmentally coordinated at the EU level (as in defense policy). What makes the EU's “system of differentiated integration” (Schimmelfennig et al., 2015) remarkable, is that it combines vertical differentiation with horizontal differentiation. As Hooghe and Marks (2023) remind us, differentiation is by no means unique to the EU. Yet, it combines vertical and horizontal differentiation to an extent that is atypical even for asymmetrical federal states (on asymmetrical federalism, see, e.g. Mcgarry, 2007; Zuber, 2010). For one, the internal differentiation of the EU occurs exactly in those domains of core state powers such as monetary, defense, and border policies that are usually uniformly integrated in federal states. In addition, the DI of non-member states in the EU–that is, external differentiation, established, for instance, through the EEA for the Single Market–is unparalleled in the world of federal states. While most contributions to this special issue focus on horizontal internal differentiation, De Blok and De Vries (2023) and Hix et al. (2023) also cover vertical internal differentiation and external differentiation, respectively.

Furthermore, the literature distinguishes different types or modes of horizontal DI. Based on Alexander Stubb’s (1996) early and influential typology, Schimmelfennig and Winzen (2020: 17–19) propose the labels “multi-speed”, “multi-tier” and “multi-menu” differentiation. Multispeed differentiation is differentiation by time. Here, differentiation is a transitional, temporary phenomenon, converging toward uniformity in a reasonable period. Multitier differentiation is differentiation across space. It durably distinguishes groups of states by their level of differentiation. Whereas the “core Europe” group is integrated uniformly (with only a few or minor opt outs or exemptions), the extent of differentiation increases as we move from the core toward the peripheral circles of states. Finally, multi-menu differentiation is differentiation structured by policies. In this mode, the depth of integration within each policy or policy area is roughly uniform. The participating states vary, however, from policy to policy. States “pick and choose” from the “menu” of policies, and each state puts together its own set of “courses”. There is no general convergence toward uniformity, nor is there a stable core of uniformly integrated member states. This is “Europe à la carte” in Stubb's typology (1996: 288).

Empirically, both multispeed and multitier differentiation are the relevant modes in the EU (Schimmelfennig and Winzen, 2020). The multispeed mode of differentiation is particularly prevalent in the DI of new member states and in market-related policies. In addition, however, the EU has developed a durable multitier structure, with a uniformly integrated core comprising the six founding member states, the Southern and the Eastern Eurozone member states, and a periphery comprising the non-Eurozone member states in Northern and Eastern Europe. Multitier differentiation is particularly pronounced in the integration of core state powers (Genschel and Jachtenfuchs, 2014), concerning, for instance, internal and external security policies in addition to monetary integration.

Another important distinction runs between exemptive (voluntary) and discriminatory (involuntary) differentiation (Schimmelfennig and Winzen, 2020: 16–17). In exemptive differentiation, member states negotiate opt-outs or are granted exemptions from treaty obligations. By contrast, discriminatory differentiation pertains to the temporary or conditional exclusion of member states from participation in EU policies. Prominent instances are the temporary exclusion of new member states from the freedom of movement in the internal market or the conditional exclusion from the Eurozone or the Schengen area of member states that do not meet economic and security prerequisites, respectively. This distinction is important insofar as it may not only affect the efficiency but also the perceived legitimacy of DI.

Finally, most studies of DI focus on de iure, that is, legally codified differentiation. This is distinct from de facto differentiation, resulting from the flexible and unequal implementation of, or informal cooperation beyond, legal obligations. In this special issue, we do not ignore de facto differentiation, but prefer to keep it conceptually separate from de iure DI, to be able to study the effects of de iure on de facto differentiation and for comparing DI with other forms of flexibility in EU policy making. Two important forms of de facto differentiation that we will also explore in this special issue are noncompliance (Andersen and Sitter, 2006; Sczepanski and Börzel, 2023) and informal cooperation between groups of member states, including the ad hoc participation in integrated policies of countries that have formally opted out (Adler-Nissen, 2009; Genschel et al., 2023). Moreover, flexibility in implementation (Zbíral et al., 2023) and “experimentalist governance” (Zeitlin and Rangoni, 2023) refer to forms of differentiation in policy implementation and policy development over time, which do without the legal exemption or exclusion of member states from EU policies.

Causes and conditions of differentiation

Research on DI has largely focused on the causes and conditions of differentiation (Burk and Leuffen, 2019), and it has done so mainly in a broadly intergovernmentalist framework (Schimmelfennig, 2019). That is, DI is analyzed as an outcome of intergovernmental negotiations on the European integration of policies and explained by the intergovernmental constellation of (domestically informed or constrained) preferences and bargaining power. It is commonly assumed that DI is driven by international heterogeneity. Given the requirement to achieve unanimity in treaty-making, and large majorities in EU legislation, negotiations are likely to become deadlocked if state preferences are far apart. DI offers an alternative to non-agreement by limiting integration to a group of states with sufficiently homogenous preferences and by excluding or exempting outliers. As the number of member states has risen, and integration has deepened and expanded into new policy domains, the international heterogeneity among the member states, the risk of decision making deadlock, and the demand for differentiation have increased.

Two major logics driving DI have been identified: an instrumental or capacity-driven logic, on the one hand, and a constitutional or sovereignty-driven logic, on the other (Schimmelfennig and Winzen, 2014, 2020; Winzen, 2016). Constitutional differentiation responds to the heterogeneity of principled integration preferences in the context of EU “deepening”. Member societies and governments are differentially concerned about transferring state sovereignty to the EU. Such concerns are most likely to arise in negotiations about core state powers–including internal and external security policies and monetary and fiscal policies–that generate concerns about state sovereignty. They are particularly relevant in Eurosceptic countries where (exclusive) national identities are strong. As Hooghe and Marks (2023) put it, constitutional DI is a “strategy to lower the temperature of self-determination”. As status quo-oriented states, Eurosceptic countries enjoy a favorable institutional bargaining situation in negotiations about deepening. Because agreements conferring significant additional powers to the EU require unanimity among governments and domestic ratification, governments of integration-skeptic states are able to threaten the other member states credibly with their veto or with domestic non-ratification (Tsebelis, 2002). To circumvent the veto and achieve an ambitious agreement, the integrationist states need to either grant the skeptics the right to opt out of the agreement or negotiate an intergovernmental inter se agreement outside the framework of the EU treaties (Kroll and Leuffen, 2015: 355). The constitutional logic explains the durable multitier differentiation of the EU in the domain of core state powers.

By contrast, instrumental differentiation responds to the heterogeneity of material integration preferences and integration capacity, above all in the context of EU enlargement or “widening”. Enlargement generates concerns about efficiency and redistribution. Old member states, or powerful interest groups in these states, fear economic and financial losses originating from market expansion, the redistribution of EU funds, and weak implementation capacity of new members. New members are in turn concerned about the market and regulatory pressures on domestic producers. Such concerns arise mainly regarding the EU's internal market and its flanking regulatory and redistributive policies. Poorer new member states typically have a weaker capacity to implement demanding EU regulation, make high demands on the EU's agricultural and cohesion funds, and create significant migration pressure. Old member states, therefore, seek to reduce the expected burdens of enlargement by imposing restrictions on the new member states’ access to the EU funds and labor market. They are also in a strong bargaining position to achieve these goals because new member states are normally more interested in enlargement than the old member states. However, permanent discrimination or “second-class membership” runs against fundamental norms of the EU (Bellamy, 2019). Moreover, new member states build capacity and adapt to the EU market and its regulations over time. The instrumental logic thus accounts largely for the multispeed differentiation of the EU in the domain of market-related policies.

Buying a pig in a poke? The missing “effect side” of DI

Recent research on DI has helped overcome the “semantic indigestion” diagnosed by Stubb (1996: 283), produced comprehensive datasets, and established coherent and empirically substantiated explanations of the development of DI over time and across rule-making contexts, policies, and countries. However, we still largely lack insights into the effects of DI. This shortcoming is problematic: after all, assessments of the desirability and decisions about the future development and design of DI should at least in part be based on evidence about its effects.

While several studies have looked into the institutional dynamics of DI at the EU level (Adler-Nissen, 2014; Dyson and Marcussen, 2010; Naurin and Lindahl, 2010; Schimmelfennig, 2016), we still know little about its effects on multilevel EU decision making and policy implementation, non-institutional international outcomes, such as inequality among the member states, and domestic outcomes, such as the politicization of EU issues. Moreover, the focus on intergovernmental negotiations and, to some extent, the domestic politics of treaty ratification, has largely neglected the micro and meso levels of public opinion, party competition, and interest group mediation. In particular, we lack knowledge on the public salience and assessment of DI and on the effects of DI on support for the EU. Such studies have only begun to appear in the past few years (Leuffen et al., 2022; Schraff and Schimmelfennig, 2020), but until recently suffered from the scarcity of data.

The empirical study of effects would also complement and inform the growing normative debate on DI (Bellamy and Kröger, 2017; Bellamy et al., 2022; Leuffen, 2022; Lord, 2015). Whereas normative justifications cannot be reduced to empirical findings, normative assessments are often linked to empirical claims. For instance, DI has been argued to strengthen the EU's very raison d’être, by contributing to reducing externalities (Lord, 2015), to threaten the EU's legal and institutional unity, certainty, and transparency (cf. Adler-Nissen, 2011; De Búrca and Scott, 2000), and to potentially undermine fundamental values such as equality, democracy, and the rule of law (Kelemen, 2019). Moreover, studies of public opinion help us to see whether the conditions of fairness or justice stipulated in the normative literature correspond to subjective assessments of citizens and contribute to the empirical legitimacy of DI.

In addition, when evaluating the medium- to long-term consequences of DI, we need to be open to questioning the (soft) rationalist assumption, present in much of the positive theorizing of DI, that governments choose and agree on DI as an efficient policy–and the best available option under the circumstances of pronounced international heterogeneity. This choice starts from the expectation that DI achieves its stated purposes of accommodating the diversity of member state preferences and capacities and thereby increasing the member states’ benefits from and satisfaction with European integration.

In a bounded rationality perspective (Jones, 1999), however, the decision takers mostly member state governments, may be unable or unwilling to gauge the effects of their DI choices. DI adds to the already significant complexity of the EU's multilevel and compound system of governance. Moreover, governments typically do not choose DI as part of a long-term strategic plan for designing the EU's constitutional architecture but as an ad hoc solution to overcome pressing intergovernmental negotiation gridlocks. Consequently, they focus on the short-term benefits for realizing their governmental preferences and for finding intergovernmental agreements, and risk discounting longer-term effects on the EU system, member states, and citizens. What seems expedient for individual governments may be detrimental to the EU as a whole or run against citizen preferences, and what proves advantageous in the short term may be outweighed by long-term disadvantages.

Finally, when seriously evaluating the merits of DI, alternative options of coping with heterogeneity must be considered. Is differentiation really preferable to stagnation (when uniform integrative progress is not possible) and to alternative forms of flexibility in EU governance?

Obviously, this special issue is not able to shed light on all these normative questions and empirical effects. However, it aims to take the broader research agenda on effects and alternatives, as well as efficiency and legitimacy, a step forward.

Linking choices and effects: Efficiency and legitimacy of DI

The literature has traditionally emphasized two criteria to assess the effects and merits of European (and global) governance: efficiency and (democratic) legitimacy (e.g. Scharpf, 1999). These two criteria can also be applied to differentiated EU governance. In this section, we discuss general considerations and assumptions about the efficiency and legitimacy of DI and describe the relevant arguments and findings of the contributions to the special issue. Similar to the general literature on international and multilevel governance (Dahl, 1994; Hooghe and Marks, 2016; Scharpf, 1999), we also consider trade-offs between efficiency and legitimacy.

Efficiency

The expectation of positive efficiency effects of DI rests on the assumption that, in a heterogeneous union, DI is Pareto-improving in comparison with uniform integration or non-integration. Major changes in the EU require the unanimous agreement of the member states (and domestic ratification in the case of treaty revisions). Such unanimity is more difficult to achieve, the more member states the EU has and the more diverse their situations and preferences are. Whereas national vetoes protect member states from having to adopt policies that they perceive as harmful, they also prevent other member states from introducing policy changes that would make them better off. DI allows governments to create bespoke arrangements that meet the needs, preferences, and capacities of member states better than a one-size-fits-all solution (De Vries, 2018; Schimmelfennig and Winzen, 2020). Whereas uniform integration is often more effective than DI – because member states that opt out would otherwise have contributed to the collective goods the EU produces, increased economies of scale, and reduced policy externalities between insiders and outsiders – DI is typically preferable to the status quo of no (further) integration, which constitutes the normal reversion point in EU-level negotiations.

Further along this line of reasoning, DI results in (more) homogenous subgroups of members sharing similar or at least compatible cooperation problems, preferences, and capacities. It can be assumed that these smaller and more homogenous subgroups are able to take faster decisions, make more internally consistent and ambitious policies, and find it easier to agree on further steps of integration (Jensen and Slapin, 2012). They are also likely to have a better policy implementation record, because poor fit between domestic and EU policies, conflicting preferences, and unequal capacities constitute major sources of noncompliance. Accordingly, we expect that DI: (a) facilitates agreement in the EU, (b) shortens decision-making time, (c) increases the depth of integration (for the insiders), and (d) improves the implementation of EU policies.

The general relationship between the differentiation and facilitation of agreements seems empirically plausible. The most dynamic period of integration in the history of the EU–the 1990s and early 2000s–was also a period of unprecedented differentiation. Moreover, apart from the internal market, the landmark developments of this period would not have been possible without DI. This applies to the EU's move into the integration of core state powers as well as to the EU's historic enlargement. By contrast, the effects of DI on routine policy making have not been studied systematically.

In their contribution to the special issue, Sczepanski and Börzel (2023) provide the first comprehensive analysis of the relationship between differentiation and noncompliance. They assume that both phenomena share similar root causes and hypothesize that DI should therefore mitigate noncompliance. Their findings, however, do not confirm this expectation. Whereas this result does not rule out a positive effect of differentiation on compliance–it may be, after all, that noncompliance would have been worse in the absence of differentiation–it cautions us against seeing differentiation as an effective remedy for implementation problems. Sczepanski and Börzel (2023) further make the important point that tolerating some noncompliance at the implementation stage may well be more efficient for European governance than exempting or excluding states from the policy altogether.

This point leads to the general issue that potentially positive efficiency effects of DI may be counteracted by the “wrong” differentiation or surpassed by negative effects. First, there are potential trade-offs between the criteria for efficient DI. For instance, maximizing the homogeneity of member states runs the danger of creating groups that are too small to meet the policy goals, produce economies of scale, and pool resources. Coordinating relations between a high number of internally homogenous groups also produces transaction costs that may diminish the benefits of DI significantly.

Second, the member states may create suboptimal groups for political reasons. Some member states may opt out of deeper integration out of general concerns about national sovereignty and identity, even though it would have been functionally efficient to join or remain with the insider group. To compensate for the resulting inefficiency of these “functionally false negatives”, insiders and outsiders then seek informal or ad hoc arrangements that straddle the legal divide but typically increase transaction costs and underprovide collective goods. The Danish opt-outs from the European Monetary Union or Justice and Home Affairs arguably fit this pattern (Migliorati, 2022). In other cases, member states were admitted to the insider group in spite of capacity deficits. The accession of some Southern European member states to the Eurozone–and the manipulated Greece entry in particular–are cases in point. Such “functionally false positives” reduce the homogeneity of insiders and, in turn, the envisaged gains in policy-making efficiency.

In their contribution, Genschel et al. (2023) show, however, that suboptimal or dysfunctional legal differentiation is not the end of the story. Treaty-based legal opt-outs can be circumvented or complemented by alternative agreements or practical steps that mitigate their negative effects on policy efficiency. This is the case particularly in the domain of core state powers, in which governments opt out of integration because they are concerned about formal sovereignty or because domestic ratification failure forces them to. Genschel et al. (2023) provide numerous examples of technical, low-level, or informal reintegration. Insiders typically welcome such reintegration because they had wanted the opt-out countries to be on board in the first place; and outsiders are able to lower their functional costs of exclusion without incurring the political costs of formal integration. The resulting “organized hypocrisy” of “differentiated differentiation” helps to partly overcome the efficiency-legitimacy trade-off of DI.

Third, DI may generate unanticipated externalities between insiders and outsiders. Positive externalities of integration create benefits for outsiders, to which they do not adequately contribute. Such externalities not only reinforce the outsiders’ preference for opting out, but also potentially incentivize insiders to leave. By contrast, negative externalities create costs for, and shift burdens to, the outsiders. They thereby produce incentives for outsiders to join but may at the same time incentivize the insiders to block their accession. At any rate, major unanticipated externalities render DI inefficient, contested, and unstable. As an example of positive externalities, the investments in preserving the Eurozone during the European sovereign debt crisis also benefited banks from non-Eurozone member states but were mainly shouldered by Eurozone governments. By contrast, in the 2004 enlargement round, the unequal opening of old member state labor markets concentrated migration flows from the Eastern new members in a few destination countries. In the UK, it came to be widely perceived as a burden that subsequently fueled the “Leave” campaign.

Fourth, DI may trigger unintended dynamics leading to wider gaps between insiders and outsiders or to further splits among the insiders. From a path-dependency perspective, DI constitutes a critical juncture that puts insiders on a different integration trajectory than the outsiders and creates incentives for the insiders to integrate ever more deeply, leaving the outsiders behind. Alternatively, differentiation may trigger “opt-out cascades” (Jensen and Slapin, 2012). The absence of the most Eurosceptic member states in the insider group may encourage the most integrationist members to push for even deeper integration against the preferences of the less integrationist insiders. And the existence of an outsider group may make it easier for these states to leave the insider group.

Finally, there is the possibility that DI reduces the chances of subsequent uniform integration when it would be beneficial to both insiders and outsiders. The institutional differentiation may translate into in-group and out-group identities over time, which put in question the common EU identity and generate a certain alienation between insider and outsider societies. They may undermine both the solidarity across groups of member states and the habit of seeking consensus and common solutions. This problem arises mostly in multitier differentiation, which creates a durable core-periphery cleavage.

Such unintended dynamics may be particularly prevalent during crises of integration, which have characterized the recent history of the EU. In their comparison of the Eurozone and migration crisis, Schimmelfennig and Winzen (2023) argue, however, that differentiated policy areas are neither more crisis-prone than uniformly integrated ones, nor as such likely to cause further fragmentation under crisis conditions. Rather, the outcomes of the two crises confirm the general mutually supportive relationship between integration and differentiation: whereas the leap in integration resulting from the Eurozone crisis has triggered, and been facilitated by, further differentiation, the absence of major reforms and the absence of further differentiation went hand in hand in the migration crisis. At the same time, DI in the Eurozone crisis followed the existing division between insiders and outsiders in a path-dependent way.

The potential or actual negative effects of DI–group divisions undermining effective policy making, creating unanticipated externalities, and producing durable and widening cleavages among the membership–raise the question of alternatives to DI that address the same issues of heterogeneity without creating formal divisions between member states. Two contributions to the special issues explore such alternatives: experimentalist governance and flexibility in implementation. In contrast to DI, both alternatives start from rules that are legally valid for all member states. Whereas experimentalist governance introduces flexibility through a recursive process of rule setting and revision, flexibility in implementation accords the member states discretion in rule implementation. In their case studies on EU electricity policy and banking union, Zeitlin and Rangoni (2023) argue that regulation under conditions of high international interdependence and uncertainty is well served by “diachronic experimentalism”, a combination of flexibility in implementation and deliberative review processes leading to regular rule revisions. Based on a novel dataset of directives in all areas of EU policy making, Zbíral et al. (2023) show that flexibility in implementation is applied primarily for legislation requiring substantial domestic policy change and in the domain of core state powers. It is also clear, however, that these two flexibility arrangements are not strict alternatives to DI–as evidenced by the fact that the banking union combines DI and experimentalist governance and that the use of DI and flexibility in the implementation of directives is positively correlated (Princen et al., 2022).

In sum, several contributions to the special issue help to put DI integration in a larger context of European governance (Genschel et al., 2023; Sczepanski and Börzel, 2023; Zbíral et al., 2023; Zeitlin and Rangoni, 2023). For one, they point out that noncompliance, experimentalist governance, flexibility in implementation, and partial reintegration pursue the same quest for efficient European governance under heterogeneity as DI. In addition, however, they do so at different stages of the policy making process and in different ways. Whereas DI is a public and rigid form of differentiation, designed to facilitate agreement at the decision-making stage, the alternatives are situated at the implementation and elaboration stages, less transparent, and more malleable. Correspondingly, DI is more likely to address highly politicized heterogeneities, whereas the alternatives tend to be used in technical settings, on practical and secondary issues and below the public radar.

Legitimacy

The effects of DI on legitimacy might be equally complex. When we refer to legitimacy here, we move beyond “output legitimacy” (Scharpf, 1999), which is closely linked to efficiency in that it focuses on (perceptions of) the welfare benefits. Rather our focus is on procedural justice and legitimacy, which build on the perception that authoritative decisions are made according to the proper procedures, typically meaning “constitutionally” and “democratically” in the EU context, but also on distributive justice, which refers to concerns about the distributional outcomes of these decisions being fair.

DI is commonly viewed as a means to accommodate preference heterogeneity, providing a close fit between member states’ preferences and capacities, on the one hand, and their actual level and scope of integration, on the other. This could improve both the efficiency and the social acceptance of EU policies. Procedurally, DI respects the sovereignty of member states and accords them the right to decide whether to participate in new integrated policies or the deepening of existing policies. Against this backdrop, we would assume DI to be perceived as a way to enhance the freedom of choice of member states and as a safeguard against domination. When it comes to distributive justice, DI ought to align the integration preferences of each member state with its needs, interests, and capabilities. In other words, DI operates according to the distributional norm of “from each according to their ability, to each according to their need”. Considering this, we would expect DI to increase the perceived legitimacy of the EU.

Generally, DI chimes well with a “demoi-cratic” theory of legitimate European integration (Bellamy and Kröger, 2017), which starts from the observation that democratic identities and processes are more strongly anchored at the national rather than the supranational level, and therefore postulates that national demoi need to retain ultimate sovereignty in constitutional decisions about integration (Cheneval and Schimmelfennig, 2013; Nicolaïdis, 2013). Accordingly, we expect that DI: (a) enjoys the support of EU actors and citizens; (b) improves the legitimacy of the EU; (c) in particular if it results from democratic procedures.

Several contributions in our special issue examine this baseline expectation. First, in order to capture the legitimacy effect, we need to understand how to best measure preferences for DI among the public. Using new survey data, Schüssler et al. (2023) demonstrate that no single survey question exists that can effectively tap into public support for DI. Rather different types exist. At the same time, the authors point toward a number of aspects of DI that find strong public support. For instance, many citizens seem to appreciate the openness and inclusiveness in the design of DI. Moreover, when evaluating DI, citizens to varying degrees care about both: the expected consequences of different DI tools for the EU and for the member states as well as their autonomy.

The contribution of De Blok and De Vries (2023) underscores the difficult relationship between DI and legitimacy. The authors show that due to the low salience of DI in domestic politics generally, ordinary people use ideological benchmarks when forming opinions about DI which mostly relate to their existing support for the EU. While pro-EU citizens are more in favor of DI when it comes to multiple speeds, as this might pose a solution to overcome gridlock, Eurosceptic citizens display more support for opt outs as a means to accommodate concerns about national identity and control. These findings suggest not only that DI means different things to different people, but that it might even entrench existing fault lines.

Moreover, the legitimacy of DI is likely to depend in part on how it is decided and implemented. For one, not all differentiations are voluntary choices. It is here that the distinction between exemptive (or voluntary) and discriminatory (or involuntary) differentiation comes into play. A decision by the insiders to prevent a willing member state from participation, usually on the grounds that it lacks the capacity to participate effectively in the integrated policy, is likely to be perceived as an act of discrimination by the excluded state. Vergioglou and Hegewald (2023) provide support for the empirical relevance of these considerations. They show that EU support increases in member states which are granted an opt-out as well as in member states that are allowed to integrate into a policy area that previously excluded them. By contrast, not being granted an opt-out or being excluded from integration significantly lowers EU support.

The institutional aspect of how DI comes about also matters for legitimacy. For example, DI legitimacy might be higher if it results from (direct) democratic procedures. In Denmark where opt-outs from EU policies have generally resulted from popular referendums, these referendums not only appear to restore a sense of political efficacy to the Eurosceptic winners of these referendums (Schraff and Schimmelfennig, 2020) but are also considered a fair procedure by the pro-integration losers (Buch and Hansen, 2002).

However, EU citizens do not have equal access to referendums on European integration and therefore do not have equal chances to assert their preferences for opt-outs against the wishes of their government (Cheneval and Schimmelfennig, 2013). Whereas in Denmark and Ireland, major treaty change is routinely put to ratification by referendum, other governments use referendums opportunistically, and some countries (such as Germany) do not allow for referendums on treaty ratification at all. Consequently, only citizens that are offered the opportunity to vote on European integration are likely to improve their perceptions of the EU's procedural legitimacy. Moreover, if citizens deprived of a direct say learn or observe that others possess this right, their perception of democratic legitimacy and their support for the EU tend to deteriorate (Schraff, 2022). In a demoi-cratic system, transnational procedural fairness thus matters in addition to national autonomy.

In addition, DI may also be perceived as unfair in the distributive dimension if member states use opt-outs to eschew solidarity and burden-sharing. The refusal of Eastern member states to implement the decision to redistribute a limited number of asylum-seekers across the EU, and their insistence on “flexible solidarity” in the deliberations about a reform of the common asylum system illustrate such uses of DI, which are likely to be perceived as unfair by the other member states. In line with this reasoning, past research has shown that citizens in Southern Europe–who have had a particularly strong demand for European solidarity during the Eurozone and migration crises–take a far more critical stance on DI than citizens in the Northern and Eastern parts (Leuffen et al., 2020).

Moreover, DI may lead to frictions or inconsistencies at the level of institutional design. For instance, Heermann and Leuffen (2020) argue that the inclusion of members of the European Parliament (MEPs) elected in countries that are not part of the Eurozone, in voting on Eurozone policies, violates principles of democratic legitimacy. In particular, these authors argue that the current practice of all MEPs voting on legislation related to DI violates the “all-subjected principle” (Gould, 2018), at least as “long as there is no uniform electoral law with transnational party lists” (Heermann and Leuffen, 2020: 1028).

One of the contributions in this special issue highlights another institutional side effect on representation. Winzen (2023) turns to the consequences of DI on EU governance by asking whether DI weakens the involvement of national parliaments in EU policy making. He shows that voluntary and comprehensive differentiation depresses parliamentary involvement to very low levels, whereas involuntary and limited differentiation does not. The author interprets this as a useful parliamentary response to the variable domestic relevance of EU policies, but that it also constitutes a challenge for the EU, as differentiation fosters differences between the member states with ramifications for the EU legal order.

Finally, it is important to consider the legitimacy of external differentiation. The article by Hix et al. (2023) examines the complexity surrounding the legitimacy of Brexit as a form of external differentiation. Specifically, it focuses on the preferences of the British public concerning the longer-term relationship with the EU. The authors show that the most preferred option among British voters is a “softer” form of Brexit: where the UK applies EU regulatory standards in return for quota-free and tariff-free access to the Single Market. However, the most preferred viable option among “Leave” voters is a much “harder” trade-off of regulatory sovereignty but limitations on UK exports. The authors highlight how difficult it is to reconcile differences between Remainers and Leavers and how this might also affect the negotiations about the precise contours of the relationship between the EU and UK in the future.

Overall, the legitimacy of DI is complex and crucially depends on the ways in which it is decided and implemented. Next to the important insights about the consequences of DI on efficiency and legitimacy as provided by the contributions to this special issue, it is important to consider possible trade-offs between the two. Whereas the exclusion of willing member states with capacity deficits may increase the efficiency of integration, it could undermine its legitimacy in the affected states. On the other hand, the inclusion of such states to increase legitimacy may negatively affect DI efficiency. In addition, democratically decided opt-outs from integration typically enhance the legitimacy of integration but often reduce its efficiency. In turn, arrangements designed to overcome these inefficiencies (such as external differentiation arrangements or informal and ad hoc cooperation with the insiders; see Genschel et al., 2023) risk being accused of domination, integration by stealth, or disregard for the people's will.

Conclusion

Our stocktaking of the literature on DI shows significant advances in this field of research, in particular, with regard to the causes and manifestations of different types of differentiation. However, we also diagnose persistent shortcomings regarding the study of the consequences of DI. This lack of scholarly attention is regrettable, not in the least because knowledge about the effects of DI is needed for formulating policy and polity recommendations on this pertinent issue of European integration. At the end of the day, only if the medium- and long-term benefits and costs on various dimensions of the EU's multilevel system of governance can be estimated, serious evaluations of DI are possible. In addition, the study of DI and its consequences in European governance would benefit from systematic comparison with differentiation in national states (Hooghe and Marks, 2023).

This special issue contributes to this endeavor by examining the effects of DI both on the efficiency and the legitimacy of the EU and addressing several salient questions. Does DI put the EU on a slippery slope of integration crises and cascading differentiation? Which governance implications does DI entail when it comes to the interactions of national parliamentarians and the quality of implementation? And how do citizens from different member and non-member states evaluate different types of DI? Do they fear discrimination, or do they see it as a chance to reconcile different preferences on European integration? The articles in the special issue provide conditional answers to these questions, which are ultimately linked to the challenge of reconciling unity and diversity in the EU and beyond. We hope that they will boost the search for theoretically sound and empirically solid answers to this challenge.