Three channels of democratic input

Like most systems that combine various territorial layers, the political system of the EU offers its citizens more than one institutionalized procedure for democratic participation. Three main channels of democratic input are to be distinguished. Each operates through specific EU bodies and is structured according to a specific logic of representation.

(a) The most obvious channel is constituted by the directly elected EP, which is explicitly conceptualized as an institution to represent European citizens at the EU level. This can be called the supranational channel of EU democracy. The Parliament’s power in legislative procedures has grown considerably with recent treaty reforms; the Lisbon Treaty that is currently up for ratification proposes yet another significant expansion. In most fields of EU policy-making, the EP can now participate through the codecision procedure in which it is a fully co-equal legislator (together with the Council of Ministers), and in which no EU legislation can be passed against its will.

There are three respects in which the EP’s powers still fall short compared to those of national parliaments. First, the EP cannot formally initiate EU legislation; the ‘monopoly of initiative’ continues to reside with the European Commission. Second, there remain a number of policy fields in which the EP has no full legislative powers, but participates in a less influential role (e.g., according to the consultation procedure, in which it can be overruled by the Council). This is presently the case in the Common Agricultural Policy, but here the Lisbon Treaty proposes a shift to codecision. No change is proposed, on the other hand, for crucial policies such as tax harmonization, police cooperation in criminal matters, and foreign policy – in these areas, limited parliamentary influence will remain the norm. Third, the EP has only limited competencies in electing and controlling the European Commission. While its assent is required for a new Commission to take office, it cannot influence the nomination of individual Commissioners and can only vote a Commission out of office with a majority of two thirds of its members.

Yet, these limitations of parliamentary power are not the only problem facing this channel of democratic input. A second problem – one that is difficult to remedy through institutional reforms – stems from the character of EP elections. In these elections, national parties run on largely national platforms, campaigns are dominated by issues of domestic politics, and the voters – if they participate at all – tend to see the elections primarily as an opportunity to sanction unpopular national governments. This led Karlheinz Reif and Hermann Schmitt (Reference Reif and Schmitt1980) to characterize the first direct EP elections in 1979 as ‘second order elections’, a characterization that remains true today (Franklin, Reference Franklin2006; LeDuc, Reference LeDuc2007). The crucial implication is that Members of the EP (MEPs) are elected on the basis of considerations that have little to do with their role as co-legislators at the EU level. It is questionable, against this background, whether the EP can adequately fulfill its role of representing Europe’s citizens in EU decision-making.

(b) The second channel of democratic input in the EU runs from national elections, national parliaments, and national governments to the Council of Ministers, which is an even more powerful decision maker at the EU level than the EP. We can call this the intergovernmental channel of EU democracy. It could be argued that this mechanism of member-state representation secures the democratic participation of the citizens, in spite of the EP’s functional deficiencies.

This second channel of democratic input is not without its problems either. First, EU-related topics hardly ever play a significant role in national elections, nor are they prominently discussed in the mass media. This means that national democratic processes do not provide good guidance for a government’s position in the Council. Second, even in controversies that do engage the citizens, national governments retain a large room for maneuver in Council decision-making. One reason for this is that national parliaments, the core institutions of member-state democracy, are often unable to keep track of EU policy-making processes; a second reason is that they hesitate to tie their own government to an overly restrictive mandate, for fear of limiting its ability to negotiate effectively in the Council (Auel and Benz, Reference Auel and Benz2007; Raunio, Reference Raunio2007). When the outcome of these negotiations does not satisfy important domestic constituencies, it is always possible to shift blame to other European governments; after all, decisions in the Council are usually made through qualified majority voting (QMV). The channel of member-state representation, therefore, is also deficient when it comes to guaranteeing that citizen preferences are respected in EU decisions.

(c) In addition to the EP and national political processes, a third channel of democratic input is constituted by European civil society, which is actively included in EU decision-making procedures, particularly by the European Commission. This is the transnational channel of EU democracy. It is best defined as supplementary to the other channels, as its procedures lack their formality and bindingness. Nevertheless, a number of authors see a considerable democratic potential in its mechanisms of policy-specific, functional representation, which seek to identify and consult relevant stakeholders in early stages of the decision-making process (Ruzza, Reference Ruzza2004; Greenwood, Reference Greenwood2007; Steffek et al., Reference Steffek, Kissling and Nanz2007).

As with the other channels of democratic input, however, doubts arise as to whether civil society participation can ultimately secure an unbiased connection between the preferences of citizens and the outcome of EU decision-making (Kohler-Koch, Reference Kohler-Koch2008). First, the stakeholder representatives consulted by the Commission are usually professional lobbyists, whose positions need not reflect the preferences that exist in society. Second, it is ultimately up to the Commission how it reacts to the positions voiced in the consultative process, and there is little the consulted groups can do if the Commission fails to listen to them.

A democratic deficit? Defining the relevant criteria

We can conclude that each of the three channels of European democracy is faced with specific problems. The question in evaluating the democratic quality of multilevel governance in the EU is whether their interaction guarantees a satisfactory measure of democracy. Obviously, the answer depends on the normative criteria by which democracy is defined. In current academic discussions, arguments against the thesis of a ‘democratic deficit’ can be found most often in two strands of democratic theory.

The first are self-described ‘realist’ approaches, which in the case of the EU are represented most prominently by Giandomenico Majone (Reference Majone1998) and Andrew Moravcsik (Reference Moravcsik2002). For these authors, the most crucial criterion for democracy is the existence of institutions that prevent the abuse of power by special interests. Judged by this standard, the EU can be considered democratic; after all, its consensual forms of decision-making and the large number of veto players ensure that all power holders are kept in tight check.

The second group of authors who often come to positive conclusions about the EU’s democratic quality are certain theorists of deliberative democracy, such as Oliver Gerstenberg and Charles Sabel (Reference Gerstenberg and Sabel2002), Hans-Jörg Trenz and Klaus Eder (Reference Trenz and Eder2004), or Gerard Delanty (Reference Delanty2007). This might surprise at first glance, since the idea of deliberative democracy, based on the work of Jürgen Habermas, is often considered a particularly demanding conception of democracy. However, many applications of this idea to the EU interpret deliberative democracy in a relatively undemanding way. Democracy is held to exist if there are procedures which ensure that all relevant arguments for, or against, a certain decision can be exchanged and assessed; by contrast, it is seen as secondary whether all citizens can actually take part in these deliberations. Viewed from this perspective, the EU is democratic because its decisions are shaped by a large number of actors – the EU bureaucracy, national governments, national and European parliaments, interest groups, etc. – which bring a variety of different perspectives to the table.

On closer inspection, both of these positions are unconvincing (Føllesdal and Hix, Reference Føllesdal and Hix2006). Each of them is based selectively on only one core criterion (control of power or exchange of arguments) and, what is more, these criteria could in principle be fulfilled in a system of ‘enlightened despotism’ as well. In other words, the above positions seem to neglect the essence of democracy, the fact that the people should govern. It is clear that in large-scale political systems, the people can be directly involved in government only in exceptional cases, while most decisions will have to be made by representatives. But representative procedures need to fulfill a number of conditions.

To define them, we can start from Abraham Lincoln’s famous triad of government of the people, by the people, and for the people, which was first introduced to EU studies by Fritz W. Scharpf (Reference Scharpf1999: 6–8, Reference Scharpf2000: 102–105). Yet, while Scharpf focuses primarily on government by and for the people – standards which in his conception are synonymous with input and output legitimacy – we argue that it is ultimately the difficulty of fulfilling Lincoln’s first requirement, government of the people, that is at the heart of the EU’s democratic problems.

1. 1. Government of the people requires that the people who are affected by political decisions (decision takers) are at the same time decision makers in a democratic system. This condition encompasses two forms of congruence: First, the people who are affected by a political development must be identical with the electorate in the selection of representatives who can control this development (Zürn, Reference Zürn2000: 188–190). Otherwise, they are ultimately ruled by powers beyond their control. Second, the people who select representatives must be able to conceive of themselves as a relevant political community capable of collective will formation (Offe, Reference Offe2000). In other words, they must constitute a ‘demos’, an imagined community about which they are genuinely concerned, for which they are willing to become engaged, and whose other members are seen as legitimate participants in the democratic process. If this condition is not met, the selection of representatives will have little to do with the collective self-determination of the people, but is likely to be guided by particularistic concerns or considerations that are irrelevant to the decisions that representatives will have to make.

2. 2. Government by the people implies that in the process of selecting representatives, as well as in influencing decisions that these might make, every member of the political community has an equal chance to make his or her preferences heard. This criterion encompasses what Robert Dahl calls ‘control of the agenda’ (no political issue should be withdrawn from popular participation), ‘effective participation’ (adequate and equal opportunities for expressing preferences), and ‘voting equality’ (everyone’s choice at the decisive stage should count as equal) (Dahl, Reference Dahl1989: 108–118). Crucially, however, Dahl also defines the criterion of ‘enlightened understanding’: All citizens should have opportunities for discovering and validating the choice that best serves their interests (1989: 108–118). This underlines why, in addition to equal participation, deliberation can be seen as a criterion for government by the people: Procedures should exist that allow for a rational exchange of arguments that not just juxtaposes various points of view, but leads to a refinement of the participants’ preferences, ideally resulting in decisions that can be accepted by everyone.

3. 3. Government for the people refers to the fact that the quality of democratic procedures is not independent of their outcomes, measured in terms of their contribution to the citizens’ common good. As Scharpf (Reference Scharpf1999: 13–21, Reference Scharpf2000: 104) points out, such output-oriented forms of democratic legitimacy encompass both positive and negative standards: In a positive sense, democratic procedures should enable representative decision makers to effectively tackle common problems that affect the citizens. In a negative sense, output legitimacy also requires checks and balances, most importantly the representatives’ electoral accountability: To make decisions that contribute to the common good, representatives must be subject to procedures that prevent the abuse (or incompetent use) of powers, and that give the people the opportunity to sanction powers holders for failed policies, ultimately by voting them out of office and replacing them by a different set of representatives (Bovens, Reference Bovens2007).

Government of the people, by the people, for the people: assessing the EU’s democratic procedures

On the basis of this comprehensive set of criteria, we can now evaluate the democratic quality of the EU’s multilevel system. In a first step, it makes sense to discuss each of the three channels of EU democracy separately (see Figure 1), knowing that in a second step, their interaction will also have to be considered.

Figure 1

Criteria of democracy applied to the EU’s channels of democratic input.

(a) For the EP – the supranational channel of EU democracy – the outcome of the evaluation is mixed. With respect to government of the people, the EP performs quite well if we focus on the congruence of the people affected by its decisions and the electorate that installs the relevant decision makers. Since the most recent treaty reforms, MEPs are able to decide on most (though not all) issues of an EU-wide relevancy, and the electorate to which they owe their mandate is identical with the group for whom their decisions are primarily relevant, namely the EU citizens. The EP fares worse, however, if we examine the congruence of the people who select MEPs with relevant imagined communities: The ‘second order’ characteristics of EP elections show that the electorate in EP elections does not conceive of itself as a political community, hence EP elections are widely seen as irrelevant and their outcome is often shaped by domestic factors that have little to do with EU-level politics.

This fundamental problem of EP elections also affects their performance with respect to the other two democratic criteria. EP elections make a contribution to government by the people as they give all EU citizens equal chances of participation. They perform worse with respect to deliberation: While the EP’s legislative procedures are characterized by a considerable amount of deliberation, the overall deliberative quality of this channel of democratic input is hampered by the fact that in EP election campaigns, rational debates about the EU’s political priorities tend to be crowded out by issues that are only tangentially relevant to the decisions that MEPs will later have to make. The EP’s deliberations, therefore, remain an elite affair that contributes little to the ‘enlightened understanding’ of the citizens.

With respect to government for the people, the EP’s effectiveness – its ability to shape policy – has grown considerably in recent years due to the extension of the codecision procedure (Maurer, Reference Maurer2007). On the other hand, the electoral accountability of MEPs is limited: Since they are elected on the basis of considerations that have little to do with their role in EU decision-making, MEPs are not accountable to the electorate for their performance in Strasbourg and Brussels. They have to run for re-election, but their record hardly plays any role in determining their chances of retaining their positions; nor is the electoral process likely to ‘send a message’ from citizens about their satisfaction with particular outputs their MEPs may or may not have supported.

(b) The second channel of democratic input (the intergovernmental channel) in the EU is constituted by national democratic processes that influence the composition and decision-making of the Council of Ministers. If we consider government of the people, this channel has the reverse strengths and weaknesses of the supranational channel (the EP). At the member state level, it can generally be assumed that the electorate views itself as a political community; national elections allow this political community to influence, albeit indirectly, the formation of Council delegations. However, in areas where unanimity is not required for Council decisions (i.e., where QMV applies), the congruence between the people affected by Council decisions and the electorate choosing the decision makers is no longer secured. In this case, EU citizens from states on the losing side of a decision are subjected to the rule by decision makers from other member states, over whose selection they have no influence.

Regarding government by the people, the Council of Ministers suffers the same weaknesses as the EP. At the national level, there are vehicles to assure citizens’ control of the political agenda, effective participation, and voting equality. However, EU-related issues play only a weak role, if any, in national election campaigns. To be sure, deliberation about EU issues does take place within national parliaments, although to varying degrees in different countries. How the national parliament concerns itself with EU issues depends on factors outside of the public’s control, such as information made available to deputies by national governments, the involvement of specialized committees, and the ability of the parliament to tie the government to a certain position (Auel and Benz, Reference Auel and Benz2007; O’Brennan and Raunio, Reference O’Brennan and Raunio2007; Raunio, Reference Raunio2007).

The final aspect to consider is government for the people. In this respect, the Lisbon Treaty – if it is ratified in all member states – will greatly increase the effectiveness of decision-making in the Council by easing requirements for decision-making through QMV. While oversize majorities would still be needed for Council decision-making, and in some areas, the unanimity requirement would remain, the Council would be less restrained in tackling problems that face Europeans. On the other hand, with the expanded use of QMV, the electoral accountability of decision makers in the Council would still be limited, for reasons discussed above. Furthermore, the leeway that national governments exercise in Council negotiations is rarely an issue in national electoral campaigns, making these contests a weak vehicle of accountability for national publics in relation to EU issues (Benz, Reference Benz2003; Auel and Benz, Reference Auel and Benz2007). Nonetheless, the Lisbon Treaty would bring some positive change in this respect, since it requires the Council to meet in public whenever it acts in a legislative function.

(c) The third channel of democratic input (the transnational one) involves civil society participation, particularly through the consultation procedures of the European Commission. Here, constituencies are defined on a policy-specific rather than a territorial basis: The Commission seeks to consult with all stakeholders who might be affected by a given policy proposal. The group of stakeholders varies from one proposal to the next, and is often difficult to define in an unambiguous way.

The basic problem that these procedures pose with respect to government of the people is obvious: While civil society representatives do have a chance to influence decision-making processes – empirical accounts vary as to the extent (e.g., Ruzza, Reference Ruzza2004; Steffek et al., Reference Steffek, Kissling and Nanz2007) – they are not the ultimate decision makers, since the Commission is not legally bound to include their suggestions in its proposals (and these proposals might be amended by the Council and Parliament). This channel of democratic input, hence, does not secure the congruence between the people affected by decisions, people selecting representatives, and the people choosing the ultimate decision-makers. It is more successful, on the other hand, with respect to the second kind of congruence: The civil society representatives who consult with EU institutions are selected by the members of their respective organizations, and these can be expected to form a distinct (if fluid) imagined community of stakeholders who care intensely about policies in a specific field.

Government by the people first of all requires the equal participation of all citizens in democratic procedures. Undeniably, this condition is not fulfilled through civil society participation. While the European Commission has made genuine efforts to identify various stakeholders and to support the organizational capacities of weak interests in its consultation procedures (Greenwood, Reference Greenwood2007), these procedures are not open to each and every citizen; it is highly unlikely that it will ever be possible to completely eradicate differences in the power resources of various interests. In short, the principle of ‘one person, one vote’ is not respected in processes of civil society participation (Greven, Reference Greven2007; Kohler-Koch, Reference Kohler-Koch2008): Some interests are difficult to organize in the first place (e.g., concerns of the unemployed), others lack adequate organizational representation at the EU level (e.g., some national minorities), and certain types of groups (such as business associations) have more clout than others. A more positive assessment seems warranted with respect to the deliberative qualities of such processes. While empirical studies have shown varying degrees of deliberation (e.g., Steffek et al., Reference Steffek, Kissling and Nanz2007), civil society participation has the clear potential of contributing to a rational exchange of arguments and the development of more broadly acceptable policy solutions – provided that all stakeholders have been identified, and no position is intentionally or unintentionally omitted.

Finally, government for the people: Existing assessments indicate that civil society participation can be remarkably effective in tapping information, correcting weaknesses in original proposals, and creating legitimacy for EU rules. The European Commission has recognized this input as a source of relevant information not otherwise readily available to its relatively small bureaucracy (Greenwood, Reference Greenwood2007). But here again, the crucial problem is accountability (Kohler-Koch, Reference Kohler-Koch2008): Neither the civil society representatives who are consulted nor the Commission can be voted out of office for their contribution to the EU legislative process.

Democratic dilemmas of multilevel governance: internal contradictions and interactions of the various procedures

What conclusion can be drawn if we sum up the individual assessments of the EU’s three democratic channels? In our view, the most important finding is that each of the three democratic criteria (government of, by, and for the people), when applied to EU multilevel governance, creates a specific dilemma that greatly complicates efforts to democratize the Union.

The first criterion, government by the people, generates what we call the congruence dilemma: It does not seem possible in the EU to square the demands for Europeanized decision-making with the configuration of imagined communities, which exist primarily at the national level. In other words, while the political developments that affect citizens in the EU member states are increasingly European in scope, the social conditions for democratic decision-making at this level remain precarious, as can be seen in the ‘second-order’ characteristics of EP elections. This constellation is indeed dilemmatic, as attempts to resolve it raise new problems: Re-nationalizing political decisions would undermine the congruence between the scope of relevant political problems (increasingly European if not global) and the electorate(s) choosing representatives to deal with them (exclusively national). On the other hand, shifting more powers to the EU would accentuate the disconnect between Europeanized decision-making and relevant imagined communities, which continue to exist primarily at the national level.

Given these difficulties, the solution found in the EU’s multilevel system can be seen as a sensible compromise: Most decisions are dependent on the concurrent approval of the EP (representing the ‘appropriate’ territorial scope) and an oversized majority of national governments in the Council (representing the ‘appropriate’ grounding in political communities relevant to the citizens); in addition, sectoral communities enjoy good access through the European Commission. The problem, however, is that the institutional linkage between these three channels of democratic input is established through non-transparent, and often informal, processes of elite negotiation (Benz, Reference Benz2003). All contentious pieces of EU legislation go through a series of inter-institutional consultations, conciliation procedures and ‘trialogues’ that remain opaque even to experts. As we shall see, this has negative implications for the other two dilemmas of multilevel governance, rendering them more problematic than they are in other democratic systems.

The internal contradiction generated by the second democratic criterion, government by the people, can be called the participation–deliberation dilemma: Giving the citizens more meaningful options for participation threatens to undermine the deliberative qualities of EU decision-making, while more deliberation implies closing the door on extensive participation. Two aspects of deliberation in particular militate against broad and equal participation (Sanders, Reference Sanders1997; Hurrelmann et al., Reference Hurrelmann, Liebsch and Nullmeier2002; for the EU, Abromeit, Reference Abromeit2002: 33–48): First, the refinement of preferences that is seen as the desired outcome of deliberation has been shown to occur most readily if participants are engaged in long-term interactions; deliberation hence works best in small groups with a fixed and stable membership, and little exposure to outside pressures. Second, the element of competition inherent in the search for the best argument privileges participants with superior material and cognitive resources; deliberation, hence, tends to result in participative inequalities.

This contradiction between participation and deliberation is not unique to multilevel governance, let alone to the EU. However, in the context of the EU’s democratic system, it develops an unusually troubling dynamic. This is largely an effect of the congruence dilemma, and the way it has been ‘resolved’ in the EU. Since the policy preferences generated through the supranational and the intergovernmental channel of democratic input are ultimately tied together in a system of elite negotiations, and the transnational channel operates through such elite mechanisms as well, democracy in the EU becomes, essentially, a two-stage affair: First a participatory, but non-deliberative process with little relevance for policy development (EP elections and national elections); then a deliberative, but non-participatory process in which policies really get made (negotiations between the EP, Council, and Commission, sometimes including stakeholders from the transnational sphere). The problem is the lack of any real connection between the two stages, which implies that the citizen’s participation tends to go nowhere, and does not have any substantive consequences for EU policy-making. But all of the standard suggestions to remedy this, for instance by extending the control of national parliaments over their governments’ actions in the Council or even by turning the Council into a directly elected second chamber modeled after the US Senate (Zürn, Reference Zürn2000: 204, 205), threaten to undermine the deliberative qualities of EU decision-making by narrowing the room for compromise and bringing to the table additional participants who are not necessarily able to contribute in an informed way to EU policy development.

A similar dilemma exists with respect to the last criterion of democracy, government for the people. We call this the effectiveness–accountability dilemma: Especially in systems of consensual decision-making, more accountability of decision makers tends to result in reduced effectiveness in reaching decisions, while more effectiveness implies less accountability. Again, this contradiction is by no means a specific trait of the EU: If representatives have to worry about their re-election (or continued support by their constituency), they will have less room for maneuver in negotiations with other decision makers, which makes it more difficult to reach an agreement. In the EU’s ‘hyper-consensus polity’ (Hix, Reference Hix2006), this general dilemma is accentuated by the fact that the three channels of democratic input all merge in a system of elite negotiations, in which mutually acceptable solutions must be found (a structure that is, as we have seen, a consequence of the congruence dilemma). In order to achieve any results, such a system presupposes not only a certain degree of secrecy, but also a significant flexibility of the decision makers – which at the same time undermines their accountability (Benz, Reference Benz2003; Auel and Benz, Reference Auel and Benz2007; Kohler-Koch, Reference Kohler-Koch2008).

A look back at Figure 1 suggests that accountability is indeed the most pressing problem of EU governance: Even when viewed in isolation, all three channels of democratic input privilege effectiveness over accountability; and their combination in elite negotiations operates according to the same logic. There might hence be a good case for reforms intended to increase accountability, but it should be clear that such reforms would at the same time reduce the EU’s decision-making effectiveness. The proposal of turning the Council into a directly elected second chamber, for instance, ignores that the EU’s system of multilevel governance differs from US federalism in that EU-level legislation is generally implemented by national executives and bureaucracies. This kind of system would generate dysfunctional results if separately elected national representatives, rather than national executives, participated in EU policy-making. Proposals for extending the control of national parliaments over the activities of their governments in the Council appear more plausible (Maurer, Reference Maurer2002), but they, likewise, cannot escape the accountability–effectiveness dilemma.

Untangling the Gordian knot: can the EU be further democratized?

Is there any way of escaping from the three dilemmas of multilevel governance, or at least attenuating their effects? Given that our analysis has highlighted the congruence dilemma as the most fundamental source of the EU’s democratic difficulties, it might appear promising to focus remedial strategies on the lack of an imagined community of Europeans. If feelings of community and identity among the EU’s citizens could be strengthened, the EP could finally emerge as a fully operational representative institution, which would make it possible to parliamentarize the Union in a more comprehensive way and do away with some of the elite negotiations that currently characterize EU multilevel governance. This, in turn, would mean that participation and deliberation could more easily be squared, and accountability could be guaranteed by electoral means.

This prospect explains why projects of identity politics remain salient in discussions about European democracy. Most importantly, the failed EU Constitution was inspired by this logic (Hurrelmann, Reference Hurrelmann2005). The very label ‘constitution’, the solemn declarations of the preamble, the articles about the Union’s goals, values, and symbols, the prominent place accorded to the Charter of Fundamental Rights – all this was intended to clarify the contours of the EU as a polity, to circumscribe the characteristics of the European people as this polity’s ‘demos’, and to define the constitutional document itself as a reference point for a European feeling of belonging.

The constitution’s failure indicates the limits of such attempts at community formation (Hurrelmann, Reference Hurrelmann2007). Large parts of the European population are not ready to accept an EU that displays the classic insignia of a state. In addition, there are enormous differences between national views of European integration, and the divergent expectations that follow from them. These differences make it impossible to define the characteristics and philosophical foundations of the Union in an unambiguous way, and to enshrine them in constitutional articles, without generating opposition in specific member states or groups of the population. Constitutional value declarations, therefore, do not constitute promising mechanisms of forging a European ‘demos’.

The same is true for an alternative strategy that focuses not on a discourse of European values and symbols, but on the increased politicization of the Union. This strategy is based on the assumption that it is possible to generate heightened interest in EU politics and strengthen Europe’s political community by increasing the salience of EP elections. This could be achieved by strengthening the link between the outcome of these elections and the partisan composition of the European Commission – either by making the Commission formally accountable to the EP, as in parliamentary systems of government, or by coming to an informal understanding that EP elections authorize the Commission to govern with a specific mandate (Hix, Reference Hix2008). But there are ample reasons for doubt whether this approach can be successful (Moravcsik, Reference Moravcsik2006). Historically, increases in the EP’s powers have not generated more interest in EP elections; rather turnout has fallen in most member states (Franklin, Reference Franklin2006). Moreover, it is difficult to see why politicization, and the shift towards majoritarianism that it implies, should necessarily generate a greater sense of community, rather than resulting in disgruntled minorities who see themselves overruled by majorities with whom they have nothing in common. Against this background, moving away from the consensual forms of decision-making guaranteed by the interconnection of three channels of democratic representation might well pose a danger to the societal acceptance for the European integration project.

We can conclude that the congruence dilemma is unlikely to be solved, at least in the short- or medium term. For the democratization of the EU, this implies that neither the intergovernmental nor the supranational channel of democratic input can conceivably legitimate EU decisions on their own. It is also difficult to imagine the transnational channel playing more than a supplementary role in EU democracy. In other words: All three channels of democratic input remain necessary. Reform proposals that suggest abandoning the EU’s distinct system of three interconnected channels offer no promising remedy to the EU’s democratic dilemmas.

Attempts to democratize the EU should hence focus primarily on the interplay of the three democratic channels, and how it can be calibrated in a way that minimizes the effects of the participation-deliberation and effectiveness-accountability dilemmas. In particular, such reform proposals should enable more meaningful participation in all stages of decision-making and introduce more relevant mechanisms of accountability. A number of options are worth being pursued in this respect:

• • Disentanglement: The idea here is that even though all three channels of democratic input remain necessary in principle, not each and every EU decision requires the simultaneous activation of all three channels. If this is so, disentanglement could reduce the democratic problems that originate in the non-transparent and elite-based mechanism of linking the three channels (Benz, Reference Benz2006: 111–113). Ideally, the EU’s policy competencies could be subdivided into three areas: (a) policy fields in which supranational decision-making by the EP is deemed acceptable even in the absence of a European political community, since issues of identity are unlikely to be involved (e.g., trade or environmental issues); (b) policy fields in which member-state representatives have to retain control because supranational majority decisions would not be accepted by the population (e.g., social policy or foreign affairs); and (c) policy fields in which decision-making requires specific technical expertise and should thus be delegated to the European Commission or independent agencies, consulting extensively with stakeholders (e.g., currency matters or food safety). This logic is not unheard of in the EU; to some extent it underlies existing variations in EU decision-making, such as different forms of EP involvement or different decision rules in the Council. It could be implemented more radically, however, by making policy formulation in each of the three areas the responsibility of only one decision maker (EP, Council, or Commission) and subject to only one channel of democratic input (supranational, intergovernmental, or transnational). This would mean, for instance, that social policy would be dealt with exclusively by the Council, environmental policy would be handled by the EP alone, and food safety standards would be set only by the Commission. In each case, the other EU institutions could play an advisory role, but would lose their ability to force policy change.

  As many policy fields cannot easily be classified as belonging to one of the above categories, and policies of different kinds are interconnected, there are obvious limits to this kind of horizontal ‘separation of powers’. Yet, even in such cases, some degree of disentanglement can be achieved by implementing a clearer functional power division that would make it possible to involve different EU institutions sequentially rather than concurrently (Benz, Reference Benz2006: 110, 111). Various kinds of legislative functions could be distinguished – such as the definition of targets (‘benchmarking’), the development of concrete policies, and the specification of criteria for implementation – and different EU institutions could be given responsibility for each of them. For example, in agricultural policy, the EP could be charged with devising policy principles (e.g., the relative importance of environmental concerns), the Council would be responsible for translating them into applicable provisions (e.g., criteria for income support to farmers), and the Commission – in cooperation with national authorities – would put these provisions into practice. This distribution of competencies could vary from one policy field to the next. All EU institutions would continue to be involved, and all channels of democratic input activated, but each would get a more clearly circumscribed mandate than is currently the case. The need for negotiations between them would thus be reduced (though not disappear entirely), which would make it possible to strengthen participation and accountability.

• • Formalization of inter-channel linkages: Even if a strategy of disentanglement were pursued, there would still be the need for joint decision-making of various EU institutions, legitimized through various democratic channels. It, therefore, makes sense to think about strategies for making their interactions more formal and transparent, which would imply that interested parts of the electorate would have more information that could be used to participate in meaningful ways and hold decision makers accountable. Unfortunately, the example of federal systems, which in many cases do not have to grapple with the congruence dilemma but nevertheless feature opaque systems of elite accommodation, is of little encouragement in this respect. Nevertheless, in addition to the reforms of the Lisbon Treaty, it might be worthwhile opening up at least the formal conciliation committees of EP and Council that are part of the codecision procedure to public scrutiny. While this would reduce the deliberative quality and effectiveness of these committees, it seems to be a reasonable price to pay for increased participation and accountability.

• • Directly-democratic mechanisms: A number of authors have suggested that direct democracy might constitute a partial remedy to the democratic dilemmas of multilevel governance (Abromeit, Reference Abromeit1998; Schmitter, Reference Schmitter2000; Papadopoulos, Reference Papadopoulos2005; Smith, Reference Smith2007). In the EU context, the creation of directly-democratic procedures would amount to the addition of a fourth channel of democratic input. This would only make sense if this channel was institutionalized not as an instrument of majoritarianism, but as a ‘negotiation device’ to allow non-elites to infuse their proposals and concerns into the EU decision-making (Papadopoulos, Reference Papadopoulos2005). The Lisbon Treaty suggests one vehicle to realize this idea, namely the citizens’ initiative. Presented as a non-binding device for requesting that the Commission make a certain proposal, it could be turned into a full-blown legislative initiative, permitting a certain quorum of citizens, after prior judicial review, to force the Commission, Council, and EP to take a stand on a proposed piece of legislation. If these institutions reject the proposal, a Europe-wide referendum would follow (Papadopoulos, Reference Papadopoulos2005). Another promising mechanism would be a directly-democratic veto of decisions that the EU’s representative institutions have taken, modeled after the facultative referendum in Switzerland (Abromeit, Reference Abromeit1998). Both types of referenda should not be decided by simple majorities, rather decision rules should mirror those used in the EU’s ‘normal’ legislative process, which would continue to function unaltered (though it would now operate in ‘the shadow of a referendum’). In this way, citizens would gain the option of participating directly and in an immediately meaningful way if a particular issue generates their sustained interest. While referenda would not have any consequences beyond the individual case involved, and the office of representative decision makers at the EU level would not be threatened by their outcome, referendum results would undoubtedly be interpreted as a signal of support or mistrust to leading politicians; this signal would also go some way towards reducing the accountability deficit.

It should be stressed that none of these proposals promises to ‘solve’ the democratic dilemmas of EU multilevel governance. All they would do is introduce some changes to the interaction of the Union’s three democratic channels, changes that would make it possible to address the most severe weaknesses of the current system: the lack of meaningful citizen participation and the lack of electoral accountability. Since it is not possible to escape from the participation-deliberation and effectiveness-accountability dilemmas, deliberation and effectiveness would necessarily suffer. But the reforms could bring about a more appropriate balance between various democratic objectives.