When and how to use differentiated integration?

Differentiated integration has advantages and disadvantages compared to alternative solutions, such as uniform integration or the status quo. When should it be chosen, and how can it be designed to yield the best possible results? The third chapter of this manual provides evidence-based advice for policymakers and practitioners for the assessment and design of differentiated integration schemes. First, it presents an evaluative framework which can help guide the assessment process. Then, it suggests the methods and sources useful to determine the comparative feasibility, effectiveness, substantive fairness, procedural fairness, and acceptance of differentiated integration schemes. Finally, it provides concrete assessment and design recommendations.

Evaluative framework

The assessment of international integration schemes is a practical endeavour relying on both positive and normative elements. On the one hand, the expected outcomes of a given scheme can in principle be identified, forecasted, and compared with viable alternatives in an objective and value-free manner, albeit with greater or lesser levels of accuracy and reliability. On the other hand, normative value judgements are required to establish the desirability of each outcome and of their overall combination: these, in turn, tend to be inherently subjective and controversial, depending on the specific interests, preferences, experiences, and sensibilities of each observer. Therefore, the role of academic research is best understood as one of assisting the decision-making process by presenting the best available empirical evidence and predictions in an accurate, clear, and neutral way, leaving to the critical judgement of each reader and to social coordination mechanisms (expert debate, public debate, the political process) the task to assess their subjective desirability and determine their actual adoption.

On the basis of a review of the relevant literature, we have developed an evaluative framework which can be usefully applied to the assessment of any kind of differentiated integration schemes or aspects thereof, their comparison with alternative options, and the design of new legal, institutional, and policy proposals (figure below). The framework includes six criteria which appear essential to the evaluation of differentiated integration schemes: feasibility, overall benefits, substantive fairness, procedural fairness, acceptance, and overall desirability. All criteria have multiple dimensions: for instance, the acceptance of a scheme may vary among governments, parliaments, electorates, supranational institutions, experts, and stakeholders. Finally, either of two principles can be applied to each criterion: the maximization of its value, leading to the pursuit of an ‘optimal’ solution, or the satisficing of its value, leading to the selection of a solution ‘good enough’ in relation to relevant alternatives and decision-making constraints (e.g. information and time available).

The use of the above-mentioned framework can assist policymakers and experts in the assessment, choice, and design of differentiated integration schemes, leading to better and more informed decisions. It is important to stress that the evaluation should always be specific (features and outcomes of a concrete differentiated integration scheme over a given timeframe), comparative (juxtaposition of relevant alternative options), and transparent (explicit and realistic assumptions about causal mechanisms, expected outcomes, and the evolution of the external context).

A number of issues should be kept in mind during this process. Firstly, a reliable objective assessment of the features and expected outcomes of a scheme is often extremely difficult: they are many and often resistant to quantification (e.g. immaterial benefits), may depend on poorly understood or causally unrelated factors, and typically depend on a correct prediction of uncertain future trends. Secondly, their normative assessment is typically highly controversial, as different individuals and interest groups are likely to hold different and sometimes diametrically opposed standpoints. Thirdly, the comparison of relevant alternatives for addressing a problem inevitably requires the selection of a handful of simplified options, but the potential variation in terms of integration models (e.g. differentiated integration, uniform integration, status quo, disintegration, experimentalist governance) and concrete design features is extremely large, and expected outcomes are very sensitive to small design variations. Therefore, a detailed expert-led assessment of differentiated integration schemes normally provides many useful indications to inform the decision-making process, but it inevitably retains a large degree of incompleteness and uncertainty and does not eliminate the need for an open democratic debate about desirable social interests, preferences, and goals.

Practical guidance on the use of the criteria is provided in the following sections.


Whenever facing a proposal for a new international scheme, establishing its legal and political feasibility is the first and most important task. This is a largely empirical criterion depending on three elements: the number of veto players who may potentially veto or delay its enactment; the number of reluctant veto players who need to be convinced or bypassed; and the amount of time and effort required to bring the enactment to its successful conclusion.

All other conditions held constant, differentiated integration tends to be more feasible than uniform integration, as it automatically reduces the number of potential participants and veto players of the scheme and may allow integrationist countries to bypass the opposition of specific reluctant ones. This effect is particularly strong for procedures relying on the unanimity of participant countries. However, both its absolute level of feasibility and the size of the feasibility gains compared to other forms of integration strongly depend on the nature of the problem addressed and on the legal instrument adopted to solve it. All major procedures are listed below, following an ordered scale ranging from least to most feasible.

The status quo always exhibits the highest level of feasibility, as any departure from it inevitably requires negotiations between countries with different interests and preferences, complex enactment and ratification procedures involving a large number of additional veto players (e.g. EU institutions, national parliaments, national electorates, national courts), and, in some instances, additional substantive and procedural legal obstacles (e.g. the need for national constitutional amendments).

The use of non-legislative means, in particular through discretionary decisions or innovative interpretations of existing law by European institutions such as the Commission, the European Central Bank, and the Court of Justice, come next. They typically only need the approval of a single supranational body, although particularly controversial decisions may be legally challenged before the Court of Justice or subsequently reversed by member states by amending the relevant legal basis. The comparative advantage of differentiated integration in this field tends to be minimal or absent, as this kind of decisions cannot change the legal applicability of the norm (uniform or differentiated) and tends to be taken by undifferentiated institutions by simple majority.

The use of inter se agreements is usually the next-best method, but not in all circumstances. On the one hand, they can be deployed in a wide range of situations (including in areas beyond the range of EU competences), can strongly reduce the number of participating states (up to a minimum of two), allow for the self-selection of participants (excluding reluctant member states), and do not require the approval of EU supranational institutions (Commission, Parliament, and Court of Justice). On the other hand, they cannot be adopted in areas of exclusive EU competence, must be compatible with existing EU law, often (but not always) involve ratifications by national parliaments and challenges before national courts, which can instead be avoided in most EU legislative procedures, and cannot rely on EU institutions to facilitate the negotiations. In general terms, differentiated inter se agreements tend to yield large feasibility gains compared to uniform inter se agreements and uniform integration through secondary EU legislation, but the latter may be equally or more feasible in situation when a consensus among national governments already exists or when decisions can be taken by qualified majority vote.

The use of secondary EU legislation usually follows. A number of special procedures facilitate the adoption of differentiated integration schemes compared to uniform ones. The enhanced cooperation procedure allows a group of minimum nine states to move forward with an integration scheme, provided that they do not violate existing EU norms, are authorized by a qualified majority of the Council (on a proposal from the Commission and with the consent of the Parliament), are launched as a ‘last resort’, and remain open for all other member states to join. Special regimes in the field of Economic and Monetary Union (EMU), Justice and Home Affairs (JHA), and defence (PESCO) automatically exclude non-members from relevant voting rights in the Council. In both cases, the support needed to enact the scheme is markedly lower than in uniform legislative procedures. EU regulations and directives approved with the ordinary legislative procedure can be indifferently used to introduce uniform or differentiated schemes. They require the support of the Commission (simple majority), national governments in the Council (qualified majority), the European Parliament (simple majority), and may be challenged before the Court of Justice (simple majority). While the opposition of few countries does not prevent the adoption of a uniform norm, the offer of full or selective differentiations can still be useful when the opposition is sufficiently large or when it threatens to lead to subsequent non-compliance problems, reprisals on unrelated issues, or a deterioration of the pro-European orientation of specific countries. Finally, legislative procedures still subject to a unanimous consent of all national governments in the Council (e.g. in taxation, social security, foreign and defence policy) are very suitable to the adoption of differentiated integration schemes, as reluctant governments may decide not to veto new initiatives if they are exempt from their obligations.

Finally, the use of primary EU legislation to amend the Treaties is the least feasible option, as it requires the unanimous consent of all national governments, domestic ratification procedures in all member states (typically a simple majority in the lower chamber, but sometimes also entailing the participation lower chambers, presidents, courts, and national electorates, the achievement of qualified majorities, and separate constitutional amendments), and additional procedural steps (e.g. a convention and intergovernmental conference in the case of the ordinary revision procedure). Most of these hurdles are time-consuming but not particularly hard to overcome: however, the need for governmental unanimity and national referendums (particularly frequent in Ireland and Denmark) have often led to substantial or unsurmountable problems. Differentiated integration can greatly facilitate the enactment of Treaty revisions compared to uniform integration, as the opposition of individual national governments and electorates can be softened by granting them permanent opt-outs from new initiatives (e.g. in the Maastricht Treaty) or by temporary limiting the rights of new EU members (e.g. in accession treaties).

Thus, depending on the goals, legal constraints, and support of a specific scheme, different methods can be used to increase its feasibility. On the one hand, legal instruments minimizing the number of veto players and the complexity of enacting procedures should be adopted whenever possible, making use of all possible margins of discretion: in particular, decision-makers should try to avoid formal revisions of the EU treaties, national referendums, and difficult domestic ratification procedures, choosing instruments with simpler procedures. On the other hand, differentiated integration often substantially increases the feasibility of a scheme compared to uniform integration. Altogether, the creation of differentiated schemes through inter se agreements, existing special regimes (EMU, JHA, PESCO), and the enhanced cooperation procedure seems particularly advantageous, while the granting of differentiations in treaty revisions yields very large comparative advantages but is remains weakly feasible in absolute terms. Finally, differentiated disintegration is rarely feasible, as requests of individual countries to regress from existing levels of integration tend to be met with fierce opposition by most other countries and supranational institutions.


The effectiveness of a scheme depends on its overall net benefits (benefits minus costs) for the entire group of its participants. Unfortunately, it is often hard to estimate them with accuracy, for two kinds of reasons. On the one hand, the objective outcomes of a scheme tend to be many, resistant to precise quantification and aggregation, sensitive to the context, and the result of a complex interplay of factors, many of them unrelated to the actual impact of the scheme. On the other hand, the subjective desirability of such outcomes may vary according to the characteristics and preferences of each interested party: for instance, the provision of the same amount of EU funds may be perceived as greatly beneficial by a poor country but as negligible by a rich country, and the achievement of a specific policy goal may be regarded as positive by observers supporting it but as negative by those opposing it. The ex ante and long-term assessment of large, complex, and politically controversial schemes is particularly problematic: for instance, economic estimates of the overall welfare benefits of past EU membership in monetary terms, leaving aside issues of immaterial gains, future developments, and subjective utility, range from 0 to 26.1 per cent of EU GDP, with a very influential paper suggesting one-off medium-term gains of 6.0 per cent but other ones indicating continuous yearly gains or no statistically significant effects. Thus, results may be more or less precise and reliable, often relying on simplifying assumptions, guesstimates, and unproven derivations from theoretical models.

The calculation of overall net benefits always depends on the specific features of a concrete scheme, its relevant alternatives, and the broader context: the nature of the policy problem addressed; the specific policy goals, solutions, and means chosen; the financial, technical, and administrative resources deployed; the effectiveness of internal decision-making, administrative, and compliance mechanisms; the future evolution of environmental trends and events; the reactions of implementing authorities, final users (European administrations,  firms, and individuals), and third parties (non-EU countries); the number and characteristics of participating countries; and the scope of the cooperation and exceptions envisaged. Nevertheless, four general categories can be distinguished, with important implications for the choice of differentiated integration.

In the first category, the net benefits of European integration can be established or assumed to be always positive and increasing with the number of participating countries: differentiated integration is here a second-best solution compared to uniform integration. In the second category, overall benefits follow a similar pattern, but individual benefits are negative for specific countries: uniform integration remains the best solution but requires compensations to win the support of the integration losers, and differentiated integration becomes more likely to be used. In the third category, net benefits are only positive for some constellations of countries and norms: differentiated integration is the way to go, tailoring the scope and membership of the scheme to approximate the ‘optimal policy area’ yielding the largest benefits. In the fourth category, the net benefits are expected to be always negative: no integration should be preferred from a functional point of view, as all other options yield negative net benefits. The bulk of the theoretical and empirical literature claims that, in general terms, most typical areas of international cooperation tend to belong to the first two categories, but more in-depth analyses regularly show the existence of entire policy areas or specific issues belonging to the third category.

Thus, it is essential for policy designers and decisionmakers to be able to correctly identify the nature of problem at hand through a concrete analysis of the potential costs and benefits of key variations in the design and membership of the scheme under discussion. Features increasing the likelihood of category 1 and 2 problems are the presence of complementary goods, positive economies of scale, high negative externalities, and homogenous needs and preferences of the participants. Typical areas belonging to these categories are the promotion of economic, personal, and cultural exchanges (e.g. the single market), the pooling of resources to solve shared problems (e.g. infrastructural investment, police cooperation), and the tackling of cross-border externalities (e.g. environmental policy). Differentiated integration is here a second-best solution yielding positive but suboptimal benefits and should be used only if uniform integration turns out to be unfeasible due to essentially political resistances among specific veto players. Features increasing the likelihood of category 3 problems are the presence of rival goods, negative economies of scale, low negative externalities, and heterogeneous needs and preferences of the participants. Typical areas belonging to these categories are those where countries have fundamentally different objective needs (e.g. fishing policy, monetary and economic policy), subjective preferences (e.g. state vs market, Atlanticism vs neutrality), and capacities (e.g. compliance with legal, administrative, and policy standards). Category 3 problems may also occur with reference to specific issues within areas otherwise suitable to broadly uniform policies, requiring an effort at selective fine-tuning. Differentiated integration is here the optimal solution, while uniform integration will yield suboptimal or even negative net benefits.

Finally, the global overall net benefits of a scheme, including the costs and benefits for third countries, should also be taken into account in the analysis. In many circumstances, a scheme may be beneficial for EU members but accompanied by high externally detrimental effects (e.g. trade diversion) or characterized by an excessively small geographical scope: here, uniform integration remains preferable to differentiated integration, but the scheme would profit from an extension beyond EU borders through EU enlargement, selective external integration, or its upgrade into a global multilateral framework (e.g. United Nations).

Substantive fairness

The substantive fairness of a scheme depends on the actual and perceived distribution of its costs and benefits as well as on a normative assessment of their appropriateness. Scholarly and popular opinions are divided on question of which kind of distribution should count as fair. At a general level, four options are possible. A non-maleficent scheme (Pareto efficiency) leaves no participant worse off and is often advocated as a minimal condition; at the same time, gains may be distributed very unequally, and truly redistributive policies may become impossible, unless benefits are framed in a highly subjective manner (e.g. the pleasure of richer countries derived from transferring money to poorer ones). An egalitarian scheme produces equal benefits for each participant, either in terms of countries or of individuals; it is often advocated as a default option but disregards the diversity in merits and needs of each participant. A merit-based scheme distributes gains according to individual efforts or contributions; however, it often reinforces pre-existing inequalities and may reward participants for accidental characteristics independent from any real merit (e.g. inherited wealth). A needs-based scheme, finally, distributes gains according to individual needs, typically from the rich to the poor; however, the latter are often criticized as underserving. At a more specific level, people may disagree about the details: for instance, too little redistribution between countries, uneven gains for sub-national territories and social groups, wrong criteria to determine efforts and needs, or an excessive burden demanded from one’s own country. The various distributive criteria tend to be mutually exclusive, although compromises can often be found around schemes contributing to a positive-sum, non-maleficent, fairly egalitarian distribution of the overall benefits of European integration (e.g. a merit-based Single Market compensated by a need-based regional policy).

As far as the objective distribution of benefits is concerned, the impact of differentiated integration naturally varies according to the specific features of the scheme, but some general tendencies can nevertheless be identified. First, differentiated integration often entails a reduction of overall net benefits but simultaneously allows individual participants to increase their relative share: the final outcome is that most countries must forfeit a certain amount of individual gains in absolute terms, but not all of them and not in the same measure. For instance, certain countries can profit from excluding direct competitors for market shares or fiscal transfers. Secondly, the benefits of European integration tend to be distributed quite unequally. One study of the economic gains of each country in 2014 from market integration and budgetary transfers (figure below), for instance, estimate them as ranging from 3.0 per cent of the GDP for the UK to 31.1 per cent for Luxembourg, with disproportional gains accruing to smaller (Benelux, Malta, Ireland) and poorer (Eastern European) countries. On the other hand, such relatively progressive redistribution between countries tends to be accompanied by a regressive redistribution within countries, with more advantaged sectors, social groups, and actors (e.g. finance, entrepreneurs, large corporations) taking the lion’s share of the benefits. The exact effects of differentiated integration are unclear but should tend to produce similar results in a more attenuated form. Thirdly, differentiated schemes requiring the unanimous consent of all EU members (e.g. Treaty opt-outs) can be typically expected to improve the non-maleficence of outcomes, as countries with negative benefits or fundamentally different preferences are more likely to opt-out. Fourthly, differentiated schemes requiring the consent of mere groups of EU members (e.g. inter se treaties or enhanced cooperations) may instead increase the maleficence and inequality of outcomes, as they can be used to monopolize the enjoyment of excludable goods and may create negative externalities for outsiders. Finally, both types of differentiated schemes may limit the scope for redistribution, as net contributors are more likely to decline participation.

As far as the choice of appropriate normative criteria and standpoints is concerned, little direct evidence exists on the normative preferences of the public about the substantive fairness of European integration schemes, their membership (uniform or differentiated), and their design features. The existing survey and experimental evidence on broader attitudes toward distributive justice and international solidarity point to a broad support for all contradictory principles (merit, needs, equality, non-maleficence), a fading of redistributive attitudes with the increase of geographical and emotional distance (stronger toward fellow citizens, weaker toward countries perceived as distant), a substantial role played by self-interest, and large variations depending on the concrete features of the scheme, its context, and the framing of the question.

Therefore, recommendations on how to assess and design substantively fair differentiated integration schemes are necessarily limited to a few methodological suggestions. First, a great effort must be made in accurately predicting the expected distribution of costs and benefits of the proposed scheme for countries and societal groups. Secondly, the creation of maleficent outcomes may be justified for willing participants but is often perceived as unfair for non-participants, as it damages them against their consent; while it is a normal feature of international relations, it should be minimized at the European level. Thirdly, the choice of the appropriate fairness criteria must be made explicitly and transparently, on the basis of the collection of direct empirical evidence about prevailing societal preferences (e.g. surveys, experiments, focus groups, or the public debate) or a discretionary choice of policymakers. Fourthly, the enacted scheme should remain open to public scrutiny and responsive to experience and changing democratic preferences. In this respect, both the use of majority-based amendment procedures (e.g. the ordinary legislative procedure) and the introduction of exit provisions seem to present major advantages over entrenched schemes which can only be revised or terminated with an unanimous consent.

Procedural fairness

The procedural fairness of a scheme depends on the actors and procedures used to enact it as well as on a normative assessment of their appropriateness.

Again, opinions are very divided on which procedures should count as fair, which many possible normative standpoints combining statist and cosmopolitan, nationalist and federalist, democratic and technocratic, supranational and intergovernmental, majoritarian and consensual, and flexible and entrenched elements. In terms of level legitimacy, people disagree on the most appropriate level of governance where decisions on specific issues should be taken. In terms of democratic legitimacy, all forms of European integration tend to enjoy a minimum level of credibility – as far as prescribed procedural norms are duly followed –, as they all ultimately rely on the consent of voters expressed in free and fair elections. However, some procedures entail a more direct and stronger democratic legitimation (e.g. those involving national and supranational parliaments elected with a high turnout) while other are more vulnerable to complaints about a substantial ‘democratic deficit’ (e.g. those exclusively involving appointed supranational institutions). In terms of democratic congruence, all forms of international and European integration involve complex compromise between many countries and are weakly responsive to subsequent changes in national and supranational preferences. Altogether, differentiated integration can be justified as an extension of national sovereignty based on the voluntary consent of national representatives, but it may be rejected by supporters of a strong sense of identity and mutual obligation between EU member stated, by non-participating countries fearing negative externalities and domination, and by people opposing the ‘constitutionalization’ of policy constraints in highly entrenched international agreements.

A body of empirical evidence on preferences about the procedural fairness of European integration schemes. Survey data on the relative strength of national and European identities, the most appropriate governance level where specific policy decisions should be made, and actual policy preferences are available in the Eurobarometer and other survey. Official turnout data and public information about the composition and internal decision-making relevant institutions can help to determine the strength and directness of democratic legitimacy. Finally, several in-depth studies have been conducted on the attitudes of specific categories of actors, such as political parties, national governments, and the general public. Altogether, these tend to suggest an improving but still inferior legitimacy of uniform European integration compared to national decision-making and the potential of differentiated integration and disintegration to satisfactorily address many substantive national concerns, but also the presence of strong objections to proposed violations of rival procedural norms (e.g. the right of reluctant countries to veto detrimental schemes or the right of integrationist countries to refuse free-riding opt-outs). If differentiation integration is chosen, schemes proposed by one’s member state or by the European Parliament, decided with a majority vote, including a large number of participants (e.g. at least 18), and where all countries ultimately join seem to be perceived as fairer than other ones. In terms of design recommendations, some general suggestions for differentiated integration schemes which might be acceptable for supporters of most normative procedural views can be offered. First, they are preferably enacted according to a restrictive understanding of proper legal procedures, avoiding clear conflicts with substantive and procedural provisions: for instance, by explicitly amending relevant provisions in the EU Treaties and in national constitutions rather than trying to get away with ‘integration by stealth’. Second, they are most suitable for areas with strong preferences for national policymaking, such as the regulation of core state powers, and strongly heterogeneous preferences. Third, procedural objections of outsider states to the formation of vanguard groups can be softened by respecting their right of veto (which allows them to block detrimental developments or to extract compensatory payments in other policy areas), by using differentiation as a ‘last resort’ after a genuine attempt at a consensual solution, by granting them some consultation and surveillance rights (e.g. non-voting participation in the Council, full participation in the European Parliament), and by keeping the ‘clubs’ open to outsiders. The enhanced cooperation procedure approximates this ideal, but a strict compliance with the first requirement would highly constrain the use of differentiation in all kinds of EU secondary legislation. Fourth, mechanisms encouraging ‘multi-speed’ differentiation through explicitly temporary provisions, clear accession criteria, and convergence policies may increase the legitimacy of differentiated integration schemes among both concerned insiders and discriminated outsiders. Fifth, they can derive a strong democratic legitimacy through the substantive involvement of national parliaments or of the European parliament in their enactment and subsequent operation: the former option is best achieved with EU and inter se treaties explicitly requiring a parliamentary ratification, while the latter one is best achieved with the ordinary legislative procedure. Sixth, they should be designed in a way which maximizes their supranational and national acceptance at the point of the enactment as well as their responsiveness to subsequent changes in preferences. The latter requirement is difficult in principle, due to the high level of entrenchment of much EU law; however, a strong political willingness of national governments to pursue acceptable compromises may help in this respect.


Finally, the empirical acceptance of a scheme refers to the degree in which it is actually supported by relevant publics, particularly national governments and national electorates. It depends both on rational reflection and on less rational considerations, such as unreflective opinions and emotions, cues from political parties, media, and peers, and unrelated contextual factors. Typically, acceptance contributes to the final decision as one of several evaluative criteria: for instance, decisionmakers may become convinced of the effectiveness and fairness of a given scheme on the basis of expert-led assessments but nevertheless discard it due to its low popular support. On the other hand, in light of its synthetic nature, high democratic status, and easy measurability, acceptance is sometimes used as a proxy for the overall desirability of the scheme, entirely replacing a more in-depth assessment: for instance, decisionmakers may wish to directly rely on the ‘wisdom of the crowd’ on issues which are normatively highly controversial, have unclear and subjective empirical outcomes, or can be better assessed by larger groups than by an individual expert.

It is perfectly possible and even relatively common to launch integration schemes with low levels of popular acceptance, provided that decisionmakers transparently justify their reasons for doing so, win the support of all relevant veto players, and are ultimately held accountable by their electorates. However, completely ignoring the preferences of the majority of citizens is hard to reconcile with many normative conceptions of democracy and representation and may subsequently lead to major practical problems.

The preferences of relevant publics on European integration and differentiation can be gauged with the analysis of the empirical evidence provided by surveys, interviews, official documents and statements, and past voting records. The InDivEU integrated database, in particular, provides direct access to a comprehensive wealth of existing and new evidence on the preferences of citizens, parties, governments, and stakeholders on differentiated integration. When choosing and designing EU integration schemes, such preferences should be explored and taken into account in the analysis.

The use of differentiated integration may either increase or decrease the support for a scheme. In a first category of situations, it softens national oppositions to further integration in reluctant countries without affecting the preferences of integrationist countries. In a second category, it elicits a strong resistance from countries preferring uniform integration and unwilling to countenance the granting of opt-outs and special treatments. In a third category, it raises the spectre of involuntary exclusion or negative externalities among laggards, leading them to prefer any other outcome (no integration or uniform integration) to the formation of a vanguard group. Aggregate and national opinions will necessarily vary according to the concrete content of the scheme, the way it is framed and presented to the relevant publics, and the broader social and political context (e.g. knowledge and politicization of European issues, approval ratings of governments, economic situation of countries).

Nevertheless, some general indications can be derived from the recent empirical evidence. With regard to integration, all countries are currently strongly supportive of EU membership and, with some isolated exceptions, of its existing uniform and differentiated policies. A plurality of countries also seems to support an overall expansion of EU powers and further integration in selected policy areas, but sizeable minorities prefer the status quo or some degree of disintegration. EU institutions and national governments are markedly more integrationist than national electorates, which not infrequently turn against further integration if a concrete scheme is properly debated and voted upon in a referendum. With regard to differentiation, opinions are more varied. The issue typically has a very low salience, with a lot of actors with unclear, neutral, or unstable preferences. Citizens and parties do not seem to oppose differentiated integration in principle, but they have many concerns about its practical details (particularly against multi-end differentiation) and probably prefer more uniform solutions. Governments, in turn, are often opposed in principle but prepared to accept concrete schemes if the alternative is no further integration. An overview of key evidence on the aggregate level of citizens’ support for uniform and differentiated integration in 2019–20 is provided in the figure below.


The assessment of differentiated integration schemes is a difficult endeavour which depends on many complex aspects: the objective features and outcomes of the scheme examined; the objective features and outcomes of relevant alternatives; the subjective interests, preferences, and normative views held by each interested actor; and the work of selection, measurement, prediction, and evaluation performed by experts. It is very sensitive to small design variations, concrete assumptions, and dominant attitudes and it can often be satisfactorily performed only ex post, when a sufficient amount of direct experience, empirical evidence, and debate about the scheme confirms or rejects theoretical expectations about its actual outcomes, advantages, and disadvantages. Nevertheless, policymakers and practitioners can rely on existing knowledge about European integration and differentiation to inform their action, thereby improving the quality of the assessment and design processes and – hopefully – leading to improved expected outcomes.

To this end, the InDivEU project has developed a catalogue of recommendations for the assessment and design of European integration schemes, summarized in the following checklist.

Altogether, differentiated integration is a useful method to accommodate a high heterogeneity in national preferences, dependencies, and capacities, but presents several weaknesses. In terms of feasibility, its use tends to greatly facilitate the adoption of any given scheme. In terms of effectiveness, it often produces sub-optimal solutions compared to uniform integration but may help maximize overall benefits in dealing with a specific category of problems, when the benefits of European integration are generally positive but will decrease with the inclusion of specific countries, due to their objective characteristics or their subjective preferences. In terms of the remaining evaluation criteria (substantive fairness, procedural fairness, and acceptance), it presents both advantages and disadvantages. Depending on its specific features and the actual distribution of aggregate and national preferences, differentiated integration may either increase or depress the legitimacy and acceptance of a given scheme. It should therefore be designed carefully, pursuing mutually acceptable compromises between integrationist and reluctant countries. In particular, the granting of concessions and opt-outs to accommodate reasonable national preferences, the avoidance of negative externalities for non-participants, and the maintenance of specific procedural rights for the latter (e.g. ability to veto or at least delay differentiation except as a ‘last resort’; consultation on its subsequent development; ability to subsequently join in a ‘multi-speed’ process) often leads to better outcomes, while differentiations perceived as threatening to widely supported values (e.g. political equality of states, legal equality of citizens, primacy of EU law), goals (e.g. solidarity, redistribution), and existing integration levels (e.g. differentiated disintegration) tend to boost opposition. In terms of legal tools, procedures relying on the unanimous consent of all states (e.g. Treaty revisions) are more acceptable to reluctant countries, procedures relying on the consent of a self-selected group (e.g. inter se treaties and existing differentiated regimes) are more acceptable to integrationist countries, and procedures relying on qualified majorities (e.g. secondary legislation, enhanced cooperation) are more suitable for compromises, as they may allow integrationist countries to forge ahead while taking into account vital interests of the reluctant ones.