Governance, Effectiveness and Legitimacy in Differentiated Integration: An Analytical Framework

ABSTRACT The mounting phenomenon of differentiated integration in the EU has hitherto been studied mainly with regard to its drivers and legal configuration. Taking the existence of differentiated integration as a given, a conceptual framework is developed for analysing its governance in practice and the conditions under which this is effective and legitimate. Referring to examples from monetary integration and Justice and Home Affairs, the framework emphasises the interplay between the legal and organisational dimensions of differentiated integration, between commitment to common policies and opportunities for participation in their development and implementation. This includes measures for assessing and conditions for explaining the effectiveness and legitimacy of differentiated integration.

After six decades of continuous deepening and widening, the European integration project is facing the triple challenge of responding to an increasingly heterogeneous membership while avoiding the creation of sharp boundaries with its neighbours and addressing an accumulating series of economic, (geo)political, environmental and health crises. Against risks of disintegration (Vollaard 2014;Webber 2014) and "postfunctionalist" stagnation (Hooghe and Marks 2009), differentiation is gaining increasing attention as a way to manage diversity in integration (see, for example, Leuffen et al. 2013;Schimmelfennig and Winzen 2020;Leruth et al. 2022). Important progress has been made retracing the historical and theoretical conceptualisation of differentiated integration, the conditions under which it emerges and the shape it takes in EU primary and secondary law. In contrast, less attention has been paid to how differentiated integration plays out in the practice of EU governance across policy areas and under what conditions this differentiated governance is effective and legitimate. We take our point of departure in the broad definition of differentiated integration (DI) as any modality of integration or cooperation that allows states (members and nonmembers) and sub-state entities to work together in non-homogeneous, flexible ways. This definition accounts for the commonly held understanding that differentiation occurs when some EU members, potentially joined by non-EU members, cooperate beyond the existing acquis communautaire, or partly opt out of it, which creates arrangements whose membership differs from formal EU membership. The decision for an opt-out can result from "instrumental" considerations, whereby members attach certain conditions to participation, which results in transitory arrangements until those conditions are met, or from a "constitutional" choice, whereby members vote against integration in a particular area for reasons of national sovereignty or identity (Schimmelfennig and Winzen 2014). Furthermore, the definition makes clear that members or actors of differentiation in the EU are not necessarily states as various sub-state actors, including regulators or local authorities, are involved in these processes. Finally, our definition is open to a crucial feature of differentiation that is widely neglected in the literature, namely that differentiated governance does not necessarily manifest itself in different levels of legal integration, but also takes place at an organisational level through nonhomogeneous participation in the various institutional venues constituting the wider European administrative space where EU-related policies are designed and implemented (Trondal 2010).
In the next section, we refine our definition of DI by examining its institutional design along the dimensions of regulatory and organisational differentiation. Drawing on insights from European Studies and International Relations (IR) theory, we then conceptualise effective differentiated governance and propose hypotheses on the possible sources of variation in its effectiveness. The third section turns to the legitimacy of differentiated governance by proposing pathways to measure and explain different levels of legitimate governance.
The purpose of this contribution is to spell out the analytical framework guiding the empirical studies in different sectors of EU policymaking conducted in the H2020 EU IDEA project 1 and partly regrouped in this Special Issue. The framework understands itself as a menu from which empirical analyses can draw inspiration without having to fully apply all aspects and can be combined with alternative measurements. The article refrains from applying the framework to empirical cases itself, but refers to illustrative examples from monetary integration and Justice and Home Affairs where appropriate.

Institutional design of differentiated governance
The governance of differentiated integration in EU sectoral policies is shaped by the nature of legal commitments on the one hand and the opportunities for organisational participation on the other hand. Fine-grained empirical categories have been proposed for measuring member states' and third countries' differentiated participation under EU primary and secondary law (Duttle et al. 2017;Leuffen et al. 2013;Schimmelfennig et al. 2015;Winzen 2016). The different quality of legal commitments and the organisational dimension of differentiation, in contrast, referring to member states' and third countries' participation in the political process from decision-shaping to decision-making and implementation/enforcement, have received less attention. Our conceptualisation therefore combines the advances made in the analysis of EU 'legal' or regulatory differentiation with approaches emphasising its interplay with the organisational dimension of differentiated governance (Lavenex 2011;2015;Lavenex, Križić and Veuthey 2021). This distinction between the regulatory and organisational dimensions allows us to grasp differentiated governance comprehensively in a procedural perspective and therefore goes beyond the analysis of differentiation as a legal outcome of political processes predominant in the literature so far.
Our analytical framework is summarised in Figure 1. We assume both the regulatory and the organisational dimension of DI governance to affect its effectiveness and legitimacy, next to other systemic factors such as issue-area characteristics, power structures and actor-specific (that is, country-specific) factors such as regulatory capacity. This section introduces analytical categories for mapping the institutional design of DI in regulatory and organisational terms before we turn to the measurement and drivers of effectiveness and legitimacy.

The regulatory dimension
The regulatory dimension of differentiation assesses the substantive scope and legal quality of a country's commitment under a given policy. The scope of regulatory commitments constitutes the main focus of existing studies of differentiated integration and refers to the extent to which a particular EU member state or, in the case of external differentiation, a third country commits to EU rules. The point of reference is the range of issues covered by a given piece of EU primary and secondary law. For comparative purposes, one can distinguish four degrees of differentiation regarding the substantive scope of regulatory commitment.
• Full commitment to the specific policy (that is, the country commits to all aspects of the policy); • Partial commitment (that is, the country commits to most aspects of the policy and only opts out of a few aspects, which hence remain under its own authority); • Punctual commitment (that is, the country has widely opted out of further integration in that policy field and commits, or opts in, only to a few aspects of the policy); • No commitment. A punctual, partial or full commitment under EU law and regulations can take different legal qualities. It can be based on supranational EU law, intergovernmental agreements or soft law, for example, through transgovernmental cooperation and the open method of coordination. 2 Even within supranational law, a large variety of commitments can be distinguished that vary in the degree to which they grant discretion to state authorities for flexible implementation (Princen et al. 2021). For the purpose of comparative analysis, we propose to differentiate the legal quality of regulatory commitments according to the mode of governance and regulation adopted in the respective policy. Here we distinguish between: • The community method, where regulatory commitment takes the quality of harmonisation via supranational law; • Intergovernmental cooperation, where regulatory commitment takes the quality of approximation based on international law; or • Transgovernmental cooperation, where regulatory commitment is based on coordination based non-legally binding DI arrangements and soft law. The Schengen cooperation in JHA is another illustrative case. Denmark, for example, is a Schengen member and participates in the free movement area, but can adopt relevant EU provisions as international law rather than EU law (thereby avoiding its direct effect and Court of Justice of the European Union jurisdiction). Ireland and the UK opted out of Schengen and retained internal border controls: at the same time, they adhered selectively and subject to unanimous agreement by 'insiders' to the flanking measures of the JHA acquis, such as asylum, police and judicial cooperation in criminal matters, and to the Schengen Information System (SIS). A more punctual possibility for opt-outs exists with the opportunity, contained in the Schengen Border Code, to reintroduce, under certain conditions and on a temporary basis, checks at the national borders. Apart from voluntary partial or full-opt-outs, the current Schengen framework also foresees an imposed partial participation for four new EU members (Bulgaria, Croatia, Cyprus and Romania) pending fulfilment of certain qualification criteria (De Somer et al. 2020). As in the Economic and Monetary Union (EMU), also in JHA member states have recurrently promoted differentiation through intergovernmental means, such as in the case of the Prüm Treaty, concluded on a German initiative with Austria, Belgium, France, Luxembourg, the Netherlands and Spain in 2005, which aimed at facilitating and widening the conditions for the exchange of data included in the various JHA databases. The Schengen framework also includes a strong external differentiated integration. Four non-EU countries are full Schengen members by way of association treaties (Norway, Iceland, Liechtenstein and Switzerland), and they also fully participate in the Dublin asylum system. While candidate countries align with the Schengen and Dublin acquis as part of their accession process, several countries of the European Neighbourhood Policy (ENP) have punctual commitments. This is for example the case with Moldova, which participates in the Eurodac system for the identification of asylum seekers via fingerprints.

The organisational dimension
The organisational dimension of DI grasps the participation of state and sub-state actors in different governance sites where differentiation plays out on an operational level. The bodies affecting the organisational dimension include the EU's formal decision-making structures (that is, the Council, the Parliament and the Commission) but also the numerous secondary organisations that make up the EU's administrative space and are involved in the preparatory or implementation phases (Trondal 2010). In contrast to the EU's central decision-making bodies, these secondary organisations are sector-specific and include competent authorities of the participating countries. They comprise, inter alia, comitology committees, 3 policy networks, EU decentralised agencies and non-EU bodies overlapping with the EU's policy, including also subnational networks such as committees of the regions or networks of local authorities (Tortola and Couperus 2022, this Special Issue). In the field of EMU, for instance, we find organisational differentiation at different levels. National central bankers of EU non-EMU countries are represented in the General Council of the ECB where they contribute to the sharing of data and information and coordinate their exchange rate policies. EU non-EMU countries also participate in the framework for reform and macroeconomic coordination in the European Semesteralthough they are subject to a looser monitoring of their fiscal policies than the members of the EMU. Still, also these countries can be sanctioned under the Stability and Growth Pact in case of non-compliance. While Euro area member states can be subject to monetary sanctions, non-Euro area members can only be subject to a suspension of the European Structural and Investment Funds. Non-Euro area member states also have voting rights in the General Council, although not for all decisions. Another important organisational venue is the Euro Summit that was established under the intergovernmental Treaty on Stability, Coordination and Governance in the Economic and Monetary Union. Non-Euro area countries are normally excluded, but they participate in discussions concerning the future of the Euro area and its fundamental rules.
The Schengen framework also bears various forms of organisational differentiation. The fully associated third countries (Iceland, Liechtenstein, Norway, Switzerland) participate in the relevant Council of Minister meetings and all relevant working groups with all rights but the right to vote. The same applies to the JHA agencies Frontex and the European Union Agency for Asylum (EUAA). Candidate and ENP countries participate on a punctual basis in these agencies, and in particular Frontex has concluded far-reaching agreements on operational cooperation in border management.
Pertinent bodies of DI can thus be located at different levels of governance (from EU to sub-national levels), and they may be involved in different stages of the policy cycle. It is therefore useful to specify and analyse the participation of actors in relevant bodies of the EU's multi-level system from the preparatory to the decisionmaking and the policy implementation phase. Agenda-setting implies "deciding what to decide" (Young 2010, 52) and allows actors to (co-)initiate legislation and policy. Decision-taking is often seen as the key aspect of political participation, but depending on whether this takes place on the basis of formal votes or consensus, also 'differentiated' participants and observers without full voting rights may have more or less influence. Decisions are then put into practice in the implementation phase, where specific bodies or networks are often in charge of translating broader policy decisions into detailed measures on the ground, as practiced for example through the EU's comitology procedure. A sensitive phase for DI is the enforcement phase which can be based on more horizontal instruments such as monitoring and transparency or occur through hierarchical, judicial means.
Obviously, this policy cycle perspective does not match one-to-one the complex realworld dynamics of policymaking in the EU. It instead serves as a heuristic tool that allows us to distinguish relevant bodies and differentiated participation therein. We focus here on formal entitlements of actors as laid down in the bodies' regulations or cooperation agreements with third countries (de jure), but of course it is important to look also at defacto participation, which may differ significantly from the formal provisions. For each body, formal participation can vary in scope as follows: • Full participation with equal rights as full members; • Partial participation (for example, as observers with full right to speak and collaborate but without the right to vote); • Punctual participation (for example, in the sense of having the right to be consulted); • No participation. In both the regulatory and organisational dimensions, differentiation implies that (some) states or sub-state entities pursue cooperation in a constellation that does not perfectly match the territory of the EU. One can distinguish between internal differentiation when EU members do not take part in cooperation arrangements adopted by other EU members and external differentiation when third countries selectively join existing EU arrangements (for example, Lavenex 2011; Schimmelfennig and Winzen 2019). In a specific policy sector, internal and external differentiation may occur simultaneously and play out in terms of legal commitments and/or organisational involvement.
In sum, a differentiated institution is hereby defined as a formally or informally organised arrangement that (i) is constructed by an internally or externally differentiated EU-related group; (ii) provides guidance in a more or less legalised way on how its members should behave in a specific sector (regulatory dimension); and (iii) establishes one or more dedicated governance bodies to sustain and move forward the cooperation arrangement (organisational dimension).

Measuring and explaining the effectiveness of differentiated governance
Differentiation is not an end in itself but rather an instrument to address common challenges by introducing a certain amount of flexibility in the complex EU policymaking apparatus. Increasing effectiveness is thus a major ambition behind the establishment of differentiated institutions. This ambition is explicit in the case of "instrumental" differentiation, which is "motivated by efficiency and distributional concerns". In this case, full participation in a given policy area is made conditional on the fulfilment of certain criteria that are seen as necessary to ensure policy efficiency (Schimmelfennig and Winzen 2014, 335). The fulfilment of certain macroeconomic criteria for joining the common currency, for example, is considered necessary for ensuring the well-functioning of the monetary union. Likewise, in the area of JHA, respect for the rule of law and effective fight against corruption are regarded as necessary for police and judicial cooperation, motivating DI under the Schengen framework. Effectiveness concerns are less imminent in the case of the second type of DI, "constitutional" differentiation, which results from a sovereign choice not to engage in a common policy (Ibid.). However, also this type of differentiation needs to be effective in order not to hamper the wider system of European governance.
We next discuss the conceptualisation and measurement of the effectiveness of differentiated institutions, and, in a second step, hypothesise the conditions under which differentiated governance is likely to be effective.

Scope
Quality Policy-shaping Policy-taking Policy implementaƟon

Benchmarks for analysing effectiveness
Empirical studies of DI have hitherto focused on mapping and measuring the phenomenon. In contrast, the effectiveness of differentiation in the EU has hardly been systematically assessed. We therefore turn to research on international regimes (for example, Underdal 2002;Young 2011) and the performance of international organisations (Gutner and Thompson 2010;Tallberg et al. 2016), which is more advanced in terms of how to conceptualise and measure the effectiveness of international institutions. At a basic level, most definitions of effectiveness draw on the "goal attainment approach", that is, they assess to what extent an organisation attains the objectives it has set for itself (Biermann and Koops 2017, 27). Yet it should be noted from the outset that different benchmarks can be used to evaluate goal attainment. Drawing and expanding on Arild Underdal (2002), we distinguish two points of reference (or standards/ benchmarks), namely "no policy change" and "ideal solution". First, the no policy change benchmark focuses on the "relative improvement" caused by the presence of a DI framework compared to "the hypothetical state of affairs" that would have occurred if the differentiated institution had not been established (7-8). 4 The approach consists in developing a counterfactual scenario on how a policy area would have evolved in the absence of the differentiated institution but with all other factors equal. Next to this 'pure' counterfactual scenario, researchers may also choose to compare the state of affairs before and after the DI framework has been put in place, in order to measure whether the newly introduced DI mechanism has been effective in terms of policymaking, policy implementation and problem-solving. When using this diachronic comparative approach, analysts need to be prudent about other factors that have also changed over time (some of these factors are presented further below).
The second point of reference (benchmark) for assessing effectiveness would consist of "some concept of a good or ideal solution" (Underdal 2002, 8) in order to capture improvement in absolute terms. Here one has to estimate beforehand what level of output, outcome or impact (see below) they expect to see in order to conclude that an institution's work was sufficient or even "ideal". Typically, such an estimation takes its point of departure in the objectives set by the protagonists of DI arrangements. However, one should be wary that objectives can change over time and that actual goals may deviate from declared ones. Also, different actors participating in DI, and in particular "insiders", that is, participants, and "outsiders", that is, non-participants, may associate different objectives with a given arrangement, making it difficult to determine a "one-size-fits-all" measurement of effectiveness (Biermann and Koops 2017, 28).
Keeping this caveat in mind, the two benchmarks can be applied to three dimensions of effectiveness, namely the effects of DI in terms of facilitating policymaking (output), policy implementation (outcome) and problem-solving (impact). With regard to (immediate) policy output, we follow Tamar Gutner and Alexander Thompson (2010, 235) and focus on "the specific tasks and narrow functions the organization is intended to perform and assess whether these are successfully carried out". Output can be measured in terms of volume, depth, coverage and reach. In the case of constitutional DI, countries that opt out of a given arrangement will measure this aspect of effectiveness with regard to the extent that they maintain autonomy in the policy at stake, whereas participants will consider DI to be conducive to output when it facilitates decision-making and cooperation (for example, Tallberg et al. 2016).
The second dimension of effectiveness, (intermediate) policy outcome, assesses whether the institution (through its policy output) affects the behaviour of target groups, that is, the scope of affected groups, and whether participating entities implement and comply with the agreed-upon output. From a longer-term perspective, the outcome dimension also entails the question of cohesion, namely to what extent differentiation creates centripetal effects over time. Do parties that were initially not associated join the framework and become integrated? This question is particularly pertinent for instrumental DI, which is designed to give outsiders "more time to adapt to EU rules and market pressures" (Schimmelfennig and Winzen 2014, 335). But from the point of view of systemic cohesion, also constitutional DI can be seen as more effective if it generates centripetal effects. JHA integration, for example, has seen a progressive de-differentiation both in the case of instrumental and constitutional DI. Most countries initially subjected to Schengen conditionality (such as Italy, Austria and Greece in the 1990s, and the Central and Eastern European member states in the 2000s) have become full Schengen members over time. An eventual failure to lift internal border controls to the remaining outsiders (Bulgaria, Croatia, Cyprus and Romania) would signal the failure of Schengen conditionality as a case of instrumental DI. Conversely, JHA has also seen the subsequent, at times patchy integration of countries that were initially opposed to cooperating, resulting in partial forms of DI such as for Denmark or Ireland (see Comte and Lavenex 2022, this Special Issue).
Finally, the effectiveness of DI can be evaluated in terms of problem-solving impact (Underdal 2002) or "macro outcomes" (Gutner and Thompson 2010). This is arguably the most intuitive form of studying effectiveness as it reflects the common definition that effectiveness is about whether an institution solves the problems that led to its creation (see for example Young 2011;Underdal 2002;Gutner and Thompson 2010). This understanding requires analysing the extent to which the overall objectives enshrined in an institution's mandate are met. In line with the benchmarks discussed above, two aspects of the impact dimension are worth considering. The first, attribution, relates to the relative improvement triggered by differentiated governance, that is, whether the existence of the institution makes any difference in terms of achieving the underlying objectives. This implies identifying whether there is a positive effect in the first place, and if so, whether it can be attributed to differentiated governance. Second, one must assess the problem-solving role of differentiation in more absolute terms, namely whether it fulfils its role of solving the underlying problem. Useful examples can again be drawn from the fields of JHA. On the one hand, it can be said that external differentiation and cooperation with candidate and ENP countries has helped reduce migration pressure on the EU. In the long term, however, the EU has become dependent on the cooperation of not always trustworthy authoritarian regimes, which now instrumentalise the issue of migration for reaping concessions (both material and immaterial, such as political support) from an increasingly vulnerable EU (Okyay et al. 2020).

Explaining variation in effectiveness
With a view to investigating under what conditions DI facilitates policymaking, policy implementation and problem-solving, and drawing on existing theories of IR and European Studies, we shall now put forward issue-related, institutionalist and powerbased explanations for DI effectiveness.

Issue-area characteristics
A first factor that might impact on the effectiveness of DI relates to the nature of the issue area (or policy field) in question. Relying on public goods theory, we here emphasise the role of excludability in a policy area, which "measures whether free-riding is possible, that is, whether or not outsiders can be excluded from the benefits of a specific good" (Kölliker 2001, 130). With regard to DI, three situations can be distinguished. First, a policy area may be characterised by low excludability, which we expect to render DI rather ineffective. If actors benefit from the problem-solving solutions provided by the DI framework even if they are not members, then they have an incentive to forego cooperation in order to free-ride on the efforts of others. Such positive externalities of DI for nonmembers are likely to be harmful for policymaking, implementation and problemsolving, because not all relevant actors would participate in the first place, and the DI mechanism as such may be unsustainable.
Second, an issue area may be characterised by high excludability, meaning the benefits of cooperation can be confined to those who participate in the DI framework. In this setting, DI is expected to be more effective because free-riding is not an option, and the DI mechanism brings together those EU members that are willing to advance on policymaking (output) and to implement the decisions that have been taken (outcome), which in turn should be conducive to problem-solving (impact). Furthermore, under conditions of high excludability from the benefits of cooperation, DI is likely to promote further integration and is thus compatible in the longer term with maintaining the EU's unity. This is because high excludability creates centripetal effects, inducing outsiders to join and implement the DI mechanism for the sake of obtaining the gains from cooperation. As pointed out by Alkuin Kölliker (2001), such centripetal effects are particularly likely to occur when the consumption of the excludable good is complementary (and not rival), in other words, when there are increasing returns to consumption.
The Schengen framework of abolished internal border controls and the adjacent SIS, a database on irregular border crossings, are typical examples of a network good exerting centripetal effects. The desire to share the benefits from this cooperation has been the major factor behind Schengen's extension from initially only five EU member states to all but the UK and Ireland (Comte and Lavenex 2022, this Special Issue).
Third, there may be situations where the role of excludability is neutral, that is, not relevant for DI. In particular, the choice for DI may be triggered by the fact that only a subset of EU members (and potentially some non-EU jurisdictions) are affected by a specific policy problem. Whether outsiders are excluded is therefore not a category of concern, because DI reflects the boundaries of interdependence that induce cooperation on an issue. This vision echoes ideas present in functionalist scholarship of European integration, namely that cooperation in the EU does not necessarily need to take place in an issue-encompassing way under a centralised EU authority but could also thrive on a task-specific basis between those actors who are interdependent regarding a specific policy. As put by Philippe Schmitter (1996, 136): Instead of one Europe with recognized and contiguous boundaries, there would be many Europes. Instead of a Eurocracy accumulating organizationally distinct but politically coordinated tasks around a single center, there could be multiple regional institutions acting autonomously to solve common problems and produce different public goods.
We expect that absent excludability concerns and negative externalities on other fields, DI has a positive effect on policymaking (output), implementation (outcome) and impact (problem-solving), because it is geared to bringing together those actors that are strongly affected by an issue within and beyond the formal EU system. These interdependent actors have the incentive to make hard choices and implement policies that favour problem-solving; in a DI framework, they are not held back by less involved (less interdependent) EU members that may lack commitment to discuss issues of low priority to them. At the same time, such an approach is not geared towards advancing EU unity. Indeed, this functionalist perspective does not even pretend that EU integration should aim for a federal-type architecture where all members transfer their policymaking competences in a homogenous way to a higher, centralised unit. Hence, a fuller assessment of the effectiveness of DI needs to address also the interlinkages between issue areas and the broader political context of integration.

Institutionalist explanations
Next to properties of the issue area, the institutional set-up of DI arrangements and participants' regulatory capacity affect their effectiveness. Here too, IR theory offers useful concepts that can be adapted to the study of DI. Institutionalist accounts in IR start from the structure of interdependence in a given issue area and, going beyond questions of excludability, derive expectations on the optimal design of international institutions. The starting point is that international institutions can help governments to overcome collective action dilemmas that hamper efforts to solve common problems through cooperation. Following on from the previous section, in the case of non-excludable public goods or common-pool resources, states have a collective interest to cooperate (for example, to drastically reduce CO 2 emissions), but each state has an interest to defect (to maintain its pattern of energy consumption) and free-ride on the efforts of others. Institutionalist scholars have argued that, under such conditions, the transfer of monitoring and sanctioning authority to an international institution can facilitate member compliance, and hence help to sustain international cooperation (for example, Keohane 1984;Simmons 1998). The expected compliance effect stems not only from the risk of suffering material penalties, but also from the loss of reputation for the infringing member.
This suggests that DI mechanisms with a strong regulatory dimension are likely to ensure member compliance with the agreed policy output, which, in turn, should enhance policy outcomes and the prospects of problem-solving. For instance, 'hard law' commitments under DI imply that in the case of non-compliance, the European Commission may initiate the infringement procedure. The latter may potentially lead to financial penalties for the non-compliant member, which in addition incurs reputational costs. To avoid such costs, DI members are expected to comply with the agreed requirements, at least more so than would be the case in the absence of hard law obligations. On the other hand, particularly complex policy areas with a high degree of uncertainty regarding optimal solutions or requiring processes of mutual learning and gradual approximation among policy-makers may benefit from regulatory arrangements based on soft law and horizontal coordination rather than hierarchical obligations (Abbott and Snidal 2000).
From the perspective of individual member states, we furthermore expect that domestic institutional capacity influences the effectiveness of DI. Scholarship on the 'regulatory state' in Europe and beyond has highlighted the role of regulatory capacity for domestic policy implementation and for states' ability to shape international rules (for example, Bach and Newman 2007;Farrell and Newman 2014;Lavenex, Serrano and Büthe 2021). Regulatory capacity is defined as "a jurisdiction's ability to formulate, monitor, and enforce a set of market rules" (Bach and Newman 2007, 831). It hinges on a sufficient level of regulatory expertise (for example, technically proficient and experienced staff as well as sufficient budgetary resources) and coherence of regulatory authority (for example, a clearly identified government unit articulating the state position on regulatory matters) (Ibid.). If such regulatory capacity is missing in DI member states, we expect negative implications for policymaking, implementation and problem-solving. This proposition is backed, for instance, by research on EU Neighbourhood Policy, where the implementation record in EU partner countries has been found to heavily depend on their domestic regulatory capacity (Börzel and Langbein 2013).
As mentioned above, instrumental DI is justified with the need for non-members to acquire requested capacities, be them linked to macro-economic performance such as in the case of the EMU or border management, rule of law and anti-corruption as in the case of JHA. Organisational features of DI arrangements can contribute to the capacity aspect of effectiveness. As discussed, DI arrangements can be more or less open to the different parties, be them 'full' participants or countries that only participate in certain aspects of the policy. Institutional arrangements that allow for a strong level of participation by non-differentiated and differentiated members and that in addition offer opportunities for learning, knowledge transfer and capacity-building should be particularly conducive to effective DI. This is because such arrangements offer the conditions for establishing a level playing field allowing each party to participate if they so wish. Research on transgovernmental relations and experimentalist governance suggests that issuespecific networks can help improve the level of capacity in the participating jurisdictions (Lavenex 2014;Rangoni and Zeitlin 2021). Transgovernmental networks, such as the European Competition Network in the field of competition law, provide a platform for domestic regulators to regularly meet and discuss relevant subjects, which facilitates the diffusion of "best practices". In such contexts, actors with initially low regulatory capacity are likely to learn from their more advanced counterparts and get socialised into common norms and practices. These insights emphasise the organisational dimension of DI. Similarly, in the field of JHA, the common curricular activities for asylum and border guards in the EU asylum (EUAA) and border control (Frontex) agencies, as well as various knowledge dissemination activities that are open to EU and non-EU member states having signed a cooperation agreement with these agencies, enhance the regulatory capacity of participating parties. By participating in various committees throughout the policy cycle, from agenda-setting to policy evaluation, regulators can learn from their counterparts, which tends to improve effectiveness in terms of policy implementation (outcome) and problem-solving (impact). We therefore suppose that differentiated institutions with a more open framework for organisational involvement show higher levels of effectiveness. Also, DI arrangements that allow for interaction with nonparticipating parties (that is, member states that have opted out of cooperation) may help to socialise the latter into the virtues of the DI framework and hence spur their willingness to implement DI output and to eventually become full DI members themselves.
On the other hand, the combination of strong regulatory commitment and low organisational involvement creates a mismatch with negative repercussions for effectiveness. Indeed, in such a DI framework, the regulatory requirements for differentiated actors may be stringent, yet these actors are not able to shape the agenda and to learn by participating in the policy process. In other words, we expect that regulatory commitment alone is insufficient to spur effective implementation by differentiated members and non-members, but that organisational commitment is equally required.

Power-based explanations
Power-based accounts suggest that international outcomes are crucially shaped by the preferences of the most powerful actors in a system (for example , Krasner 1976;Drezner 2007). Relying on their enormous market size, great powers can, for instance, use economic coercion to align recalcitrant counterparts with their policy preferences. While great powers can induce smaller counterparts to adapt their rules, coercive capabilities cancel each other out when two or more players have relatively equal economic power. Therefore, scholars have suggested that great-power agreement on common rules is required to effectively shape global rules (Drezner 2007).
Regarding the effectiveness of DI, the power-based explanation can be approached from an internal and an external perspective. From the EU internal perspective, a DI framework should require support from the most powerful member states in order to be effective. In particular, the Franco-German axis is commonly seen as the linchpin of European integration, and appears equally important for the success of DI projects. The two hitherto most consequential experiments in DI, the Schengen cooperation and the creation of the Euro, were both led by Germany and France (see also Comte and Lavenex 2022, this Special Issue). If the joint leverage of France and Germany is not enough to spur a common European approach, at least they are supposedly able to push through common policies and assure their implementation in the narrower framework of DIprovided their role is seen as legitimate by the other participating and non-participating member states (see below).
From an EU-external perspective, the success of DI may hinge on the positioning of foreign great powers such as the United States (US), Russia or China. If the preferences of these actors go against the objectives of a differentiated EU institution, they may seek to distract the process by exerting various sorts of pressure. For instance, the development of an independent European security and defence architecture has been curtailed by, among others, the US's preference to maintain NATO as the focal point of Western defence cooperation with the aim of "safeguarding direct US influence on European security affairs" (Pohl 2014, 31).

Measuring and explaining the legitimacy of differentiated governance
In order to be sustainable, DI must not only produce effective governance, it must also be considered legitimate in the eyes of the affected societies. In this section, we define and present options for measuring legitimate DI and then theorise its potential triggers.

Definition and measurement of legitimacy
Drawing on Jonas Tallberg and Michael Zürn (2019, 585), we define legitimacy as "beliefs within a given constituency or other relevant audience that a political institution's exercise of authority is appropriate". Following this definition, we approach legitimacy from a sociological perspective focusing on citizens' beliefs and perceptions, rather than reflecting on "the normative goodness of an institution" as practised by political theorists (Ibid.; Bühlmann and Kunz 2011, 319). For our purposes, this implies that we have to analyse legitimacy from the perspective of different audiences, notably citizens in EU member states that participate in a DI framework, citizens in EU member states that have opted out (constitutional DI) or are yet to fulfil the conditions for participation (instrumental DI), and citizens in differentiated non-member states (that is, non-EU members having opted in).
Existing literature and cross-national surveys provide some guidance on how to operationalise legitimacy beliefs. Most empirical scholarship relies on one of three proxies, namely questions on public 'support', 'trust' or 'confidence' in the respective institution, that is, the EU or some other global governance framework (see Dellmuth 2018; Dellmuth and Schlipphak 2019). Lisa Dellmuth and Jonas Tallberg (2018, 321), for example, measure the legitimacy of international institutions by asking "respondents to indicate their confidence [in the respective institution] on a quasi-continuous scale from no confidence at all (0) to complete confidence (10)". Some studies go further in trying to measure different aspects of legitimacy. For example, in a survey experiment, Brilé Anderson et al. (2019) asked respondents to give their opinion on several statements that were considered to "capture both deference to the decisions of a global governance institution and respondents' social affinity with the institution".

Sources of variation in legitimacy
The explanatory framework proposed by Tallberg and Zürn (2019) suggests that legitimacy depends on three features of an institution (in our case, of a differentiated EU institution): its effectiveness in terms of impact; its authority; and its procedural standards. To these, we add two more political factors linked to the origins and conditions of DI: the distinction between instrumental and constitutional DI, and the role of great powers in fostering differentiation.
First, an international organisation's legitimacy depends on its performance or impact, that is, its capacity to deliver effective problem-solving. A lack of effectiveness puts in danger "the right to rule" of international institutions in the eyes of relevant stakeholders (Buchanan and Keohane 2006, 405). As put by Allen Buchanan and Robert O. Keohane: The basic reason for states or other addressees of institutional rules to take them as binding and for individuals generally to support or at least to not interfere with the operation of these institutions is that they provide benefits that cannot otherwise be obtained. If an institution cannot effectively perform the functions invoked to justify its existence, then this insuficiency undermines its claim to the right to rule (422).
We therefore expect that differentiated institutions with higher levels of problem-solving effectiveness or impact (as discussed above) have greater legitimacy in the eyes of citizens.
At the same time, the transfer of decision-making authority to an international institution risks creating legitimacy deficits because "international authority challenges the sovereignty and autonomy of nation states -the traditional locus of legitimate authority for citizens and elites" (Tallberg and Zürn,593). Authority is hereby defined as "the recognition that an institution has the right to make decisions and interpretations within a particular area" (586). In the case of DI, the extent of authority corresponds to the regulatory dimension of our DI concept, which measures the legal scope and quality of member state commitments within the DI framework. This leads to the expectation that differentiated institutions with a stronger regulatory dimension (that is, higher authority) tend to have a lower level of legitimacy in the eyes of citizens because they impinge more strongly on state sovereignty and autonomy. It would be interesting to test whether the strong reliance on intergovernmental treaties for DI in core state powers such as monetary and JHA cooperation addressed in the first part of this article bears more legitimacy in the eyes of citizens than similar arrangements based on supranational competence.
Third, Tallberg and Zürn (2019) argue that the legitimacy of an international organisation hinges on its respect for procedural standards. One widely recognised procedural ideal in democratic systems is the right to participation of relevant actors. This procedural standard is reflected in the organisational dimension of our DI concept, which measures the participation rights of differentiated EU and non-EU members through the whole policy cycle from agenda-setting to policy implementation. This aspect of procedural participation is particularly important for instrumental DI that occurs at the request of insider countries and tends to be imposed on newcomers that have first to qualify for full membership. Opportunities for participation, however, matter also for constitutional DI, in particular when the progressive integration within the in-group creates strong externalities on the out-group. Thus, we expect, first, that differentiated institutions with greater procedural/organisational openness have higher legitimacy in the eyes of citizens from differentiated non-members and from EU members that do not fully participate in the regulatory dimension: thus, in the case of JHA cooperation, citizens from Iceland, Norway, Liechtenstein and Switzerland, whose governments sit in all relevant organisational fora, should assess the legitimacy of their DI arrangements more positively than citizens from candidate and ENP countries, whose governments are excluded from the council of ministers or agencies' management boards. Yet, such differentiated institutions should entail lower legitimacy beliefs among citizens from full DI members, because granting access rights without legal commitments may be perceived as giving an undue advantage to less committed EU members and nonmembers. Another important procedural safeguard in liberal democracies is compliance with the rule of law, transparency and human rights. It is here that external differentiation in JHA probably encounters the main legitimacy challenges. Whereas it can be said that intensive cooperation with third countries on the migratory roots has helped contain migratory flows towards the EU in the short term, this cooperation has occurred outside the procedures provided by the treaties and involves partners whose human rights records are more than problematic, thereby seriously undermining the legitimacy of the whole arrangements (Okyay et al. 2020).
To this account, we add a fourth, power-based explanatory factor that appears relevant in the EU context. As developed in the previous section, support from EU great powers is supposedly conducive to DI effectiveness; Germany and France in particular are arguably able to provide leadership by exerting leverage in DI-related policymaking and implementation. Yet, leadership by great powers may have an inverse effect on legitimacy, because it easily creates the impression of hegemonic repression. This is especially the case when the conditions for accessing instrumental DI arrangements are defined onesidedly by powerful insiders and when the criteria used to assess conformity with these conditions are not applied properly, opaque or contested.

Conclusion
Combining insights from existing research on differentiated integration, European Studies and IR, this article proposes an innovative framework for mapping and measuring differentiated integration along its regulatory and organisational dimensions and introduces a toolbox for assessing and explaining effectiveness and legitimacy. The proposed measurement of the institutional design of differentiated integration provides the basis for a comparative assessment across policy areas. This allows identifying different types of differentiation and assessing its extent. We further propose common definitions for effective and legitimate governance that draw on a number of identifiable indicators. Finally, the proposed analytical framework highlights how the institutional design of differentiated integration can affect its effectiveness and legitimacy, in addition to other variables such as domestic institutions, characteristics of the issue area, power constellations and the legitimation strategies of political actors. As differentiated integration is likely to stay and proliferate, we hope that this analytical framework will prove useful beyond the research conducted in the EU IDEA project and inspire, in full or in parts, future research on effective and legitimate differentiated governance in Europe.