Is Europe going far enough? Reflections on the EU's economic governance

Europe's economic governance is not only highly complex, but also increasingly inefficient and therefore unsustainable in the long run. This conclusion is reached from the theory of collective action and the difficulties in democratic legitimacy. The solution would be the creation of a European government accountable to European citizens.

The Lisbon European Council in 2000 designed a comprehensive and integrated strategy for all these separate processes by stipulating that Europe should be 'the most competitive and dynamic knowledge-based economy in the world, capable of sustaining economic growth with more and better jobs and greater cohesion', but progress towards its realization has been slow. Not surprisingly, the efficiency of Europe's complex governance has been questioned (Begg et al. 2003;Gros and Hobza 2001;Collignon 2001) and concern about the democratic legitimacy of existing policy procedures ('closeness to the EU's citizens') has motivated the elaboration of a European Constitution by the European Convention.
The most recent institutional innovations have strengthened intergovernmental policy co-ordination; no further policy conferrals to the European level have taken place, while new procedures like the OMC have been extolled as 'reorganising the modes of European construction' (Telò 2002). Yet, their optimality must be questioned. While economic integration requires institutions with the authority to enact Europe-wide policies, they may be traded off against costs from imposing homogeneous policies upon heterogeneous groups. Subsidiarity is supposed to be welfare increasing (Alesina and Wacziarg 1999; for a critique, Collignon 2003).
Are the newly developed methods of intergovernmental co-ordination appropriate for the tasks they are meant to accomplish? While not all European objectives need to be decided or implemented by a centralized authority, it is also clear that not all forms of intergovernmental policy co-ordination have the same effectiveness. There is a role for both intergovernmentalism and the community method. Which policy regimes are likely to produce best results?

Externalities: the rationale for policy co-ordination
There is a growing recognition that a proper functioning of EMU requires a well-developed co-ordination framework in order to ensure cohesion between the fully unified monetary policy and many other policy areas under separate national jurisdictions. Because policy preferences are defined in national constituencies, different governments have different preferences and objectives. But at the same time the growing interdependence between national economies within the same monetary framework has caused an increasing range of spillovers into other jurisdictions. Many policy variables such as inflation, interest and exchange rates but also the aggregate fiscal stance are now produced 'jointly' (von Hagen and Mundschenk 2002). What one member state does, increasingly affects all others and inconsistent policy objectives would lead to welfare lowering outcomes. Policy co-ordination is necessary when there are autonomous and non-unified decision-makers with heterogeneous preferences and objectives -and when their actions cause spillovers into each other's jurisdictions. It is therefore a feature of intergovernmentalism.
Besides the argument of interdependence calling for co-operation, we must also consider the possibility of preference change over time as a crucial element of efficient governance structures. If different actors converge on similar objectives, other forms of co-ordination are required than if preferences are disjunct and inflexible. Furthermore, if policy objectives need to be revalued in changed circumstances, discretionary forms of policy-making are more adequate than if one needs to stick to a given policy rule over time. These considerations determine different regimes of policy co-ordination.
Commonly, four arguments explain the need for co-ordination and these potential benefits apply to both supply and demand side policies: (1) National actions may spill over directly into neighbouring countries. For example, state aids may cause distortions in the single market and must therefore be controlled at the European level. (2) Indirect effects are particularly prevalent for European collective goods. Macroeconomic variables like inflation, interest or exchange rates concern all economic agents in Euroland and if independent national policy actions would cause the ECB to adjust monetary policy, the economic conditions in the whole area would be affected. (3) Co-ordination should prevent or reduce the likelihood of free-rider behaviour by member states. (4) Some argue that policy co-ordination may be useful in deflecting criticism of unpopular but necessary policy action at the national level (Drazen 2000).
Objections to policy co-ordination have focused on three aspects, mainly related to macroeconomic co-ordination: (1) The size of spillover effects and potential gains from co-operation may be small (Gros and Hobza 2001;Currie et al. 1989). For example, the aggregate demand effect of government spending may be compensated by rising interest rates, at least if the economy is close to equilibrium.
(2) Policy co-ordination may cause potential dangers to the independence of the European Central Bank (ECB) (Alesina et al. 2001;Issing 2002). (3) The observed differences in national policy preferences may require independent policies to accommodate different 'tastes' (Alesina and Wacziarg 1999). The first two objections are related to the efficient policy mix in Euroland, the third is about its optimality and is analytically distinct from efficiency considerations. I will discuss preference heterogeneity below, but the efficient policy mix is a necessary condition for the sustainability of macroeconomic stability, although different policy mixes may set different incentives for economic growth.
If co-ordination fails to produce a coherent set of actions, delegation of decision-making to a central authority may be required. The European governance framework has tackled the issue of co-ordination from four different perspectives.
(1) Efficient macroeconomic policy is to ensure that the economy achieves non-inflationary, stable growth and high employment (TEU, art. 2). This implies that policy is actually capable of delivering these collective goods. Although this claim is not undisputed, it now seems to command a reasonable consensus among economists (Allsopp and Vines 2000). In particular, there is agreement that inflation is ultimately always a monetary phenomenon and therefore the central bank needs to be independent to prevail against other actors when there are inflationary shocks. But this independence also requires that monetary policy-making is unambiguously centralized. The creation of the ECB was therefore an indispensable condition for the consistent design of monetary policy. However, European central bank independence implies a rather unique co-ordination regime for monetary and fiscal policy.
(2) Even if the central bank is independent, non-co-operative games between monetary and fiscal authorities can lead to destabilizing outcomes when the objectives between the two are not consistent (Sargent and Wallace 1981). Therefore, fiscal policy needs to be constrained. This was the economic rationale behind the SGP. By stipulating precise rules for 'avoiding excessive deficits', the Pact was meant to reduce the potentially destabilizing power of fiscal authorities and free the ECB to set interest rates compatible with macroeconomic equilibrium. This logic implies interaction between the single monetary policy and the European aggregate fiscal stance. Under Europe's present-day economic governance, this aggregate is only defined indirectly by the SGP, which stipulates one budget rule applying to all -namely, 'close to balance or in surplus over the medium term', so that the aggregate structural budget position is also in balance.
(3) Even if all national budgets were in perfect balance, inflationary pressures could still emerge from nominal wage settlements exceeding labour productivity increases. Sustained unit labour cost increases above the ECB inflation target would provoke monetary tightening with negative spillover effects via interest and exchange rates on economic growth in the short term, but also on investment and employment in the long run (Collignon 1999). The tool for internalizing these externalities was the creation of the Macroeconomic Dialogue at the Cologne European Council. It has been fairly soft, i.e. nonconstraining, partly because the ECB refused ex ante co-ordination as a matter of principle, partly because the nature of wage bargaining in Europe is so diversified that uniform procedures are neither applicable nor desirable. The Macroeconomic Dialogue works primarily through the improved flow of information clarifying the macroeconomic environment for wage bargainers.
(4) An efficient policy mix will keep aggregate supply and demand in balance. But it will also have an effect on the growth potential of the EU economy, although the impact may be limited. The transmission mechanism from stabilization policy to long-term economic growth works essentially through investment in the stock of physical and human capital. Higher growth therefore requires a macroeconomic policy mix that creates an incentive for investment, but this effect would be significantly amplified if structural reform policies created positive externalities in the form of higher productivity. In the context of the European policy framework, the BEPG are supposed to internalize these reciprocal externalities between macroeconomic management and structural reforms by formulating a coherent document that gives orientation to national policy-makers. The procedures for the co-ordination of supplyside reforms were already set by the Luxembourg and Cardiff Processes, but only Lisbon designed the overall strategy for achieving these growth and welfare improvements, with the OMC as its instrument. The OMC is now applied in a range of fields, where general guidelines or objectives at European level are translated into national action programmes (Rodrigues 2003).
Hence, the EU has both a need for and a wide range of instruments for policy co-ordination. But how are they used?

Governance regimes as forms of co-ordination
Welfare increasing policy co-ordination can be modelled as a game, where the pay-off to the players is highest when they co-ordinate their strategy. The constitutional question is which mechanisms allow independent authorities to consistently Pareto-improve welfare over time. Game theory shows that choice consistency is more easily established in sequential and repetitive games than in games with simultaneous moves, but this advantage comes at the potential cost of dynamic inconsistency with related problems of credibility (Kydland and Prescott 1977). The co-ordination between monetary and fiscal authorities can be set up as a sequential game, but intergovernmental co-ordination, like the definition of the aggregate fiscal policy stance, cannot be so.
Successful policy co-ordination has two dimensions: the consistency of objectives between autonomous partners and the dynamics in the interactions over time. If different actors make their decisions simultaneously, they may create positive or negative spillovers or externalities for others, who may anticipate them and respond by taking these decisions into account. If their strategies are mutually consistent, a Nash equilibrium exists, although the outcome is not necessarily Pareto-efficient. Policy co-ordination is then one way of improving the outcome for both. The related compliance problems can be dealt with by setting up specific incentive structures (positive rewards, negative sanctions and institutions as commitment devices), but also by convincing others of the 'rightness' of a proposed action. In the first case preferences are given and feasible outcomes are the result of strategic bargaining and side-payments. In the second case, the policy preferences actually change owing to public deliberation. The relative importance of incentives and persuasion distinguishes hard from soft co-ordination regimes. The flexibility by which policy objectives can be changed determines the degree of discretion versus rule-governed policy actions.
When decision-making is sequential, ex ante co-ordination is not necessary and can be replaced by implicit co-ordination. Actions are then co-ordinated between a leader and follower. A typical case is the interaction of fiscal and monetary policy in EMU. The ECB interprets its status of independence as incompatible with ex ante co-ordination, if a previous commitment to coordinated policy action would constrain its actions. But this does not necessarily prevent the central bank from setting interest rates at a level that is consistent with macroeconomic equilibrium. In contrast to the Sargent and Wallace (1981) framework of a 'chicken game', interactions between European fiscal and monetary policy should be modelled as a sequential game, where the ECB is the Stackelberg follower (Allsopp and Artis 2003: 14;Howarth and Loedel 2004). The sum of national budget positions first determines the aggregate fiscal stance in Euroland; while the ECB preserves the freedom to decide which monetary policy stance is compatible with it. The potential handicaps of this institutional set-up are limited by the ex ante constraint of the SGP on national budget positions and by the commitment of the ECB to a conservative attitude on fiscal policy. The sequential model of fiscal and monetary policy coordination therefore reduces the inflationary dangers. Hence, monetary and fiscal policy in EMU is only indirectly and sequentially co-ordinated, i.e. ex post, and not simultaneously or ex ante. As Issing (2002) clearly states, the ECB is operating with such a sequential model in mind.
Co-ordinating monetary and fiscal authorities is one thing, policy coordination between European governments is quite a different one. Here sequential or implicit co-ordination is not possible because the strategic interactions may resemble the prisoner's dilemma. In this case policy co-ordination by contracting and punishment for defaulting would improve the policy outcome. For example, the aggregate fiscal stance is determined by all member states simultaneously. National governments retain responsibility for national budgetary and structural policies with high potential for spillovers, but they are exclusively accountable to their domestic constituencies. Hence, there is no European institution able to provide 'implicit' leadership to others. But if fiscal rectitude is politically costly at home, although optimal for the Union, a dominant strategy would be to behave with fiscal laxity. Hence the need to establish co-ordination rules in the form of the SGP. This is probably the most prominent co-ordination problem in Europe today, but the argument also applies to a much wider range of policies.
Traditionally, the EU has used four methods to deal with the externalities caused by simultaneous decision-making of autonomous governments.
(1) Voluntary co-ordination between member states. If governments can agree in advance on what they want, policies are ex ante co-ordinated through consensus. There is then no risk of default, because each individual government does exactly what they all want collectively. Hence voluntary co-ordination requires at least coherence, if not convergence of preferences between different actors. Specific co-ordination processes serve to facilitate the emergence of a consistent policy consensus. The Macroeconomic Dialogue and the OMC are forms of voluntary co-ordination where public deliberation, the exchange of information, and peer pressure through naming and shaming facilitate preference convergence. The voluntary commitment to common objectives eliminates negative externalities and creates welfare effects. However, this form of co-ordination only applies to discretionary policy-making, as voluntary agreements are made case by case and each partner retains the full sovereignty to join or withdraw from the process. Begg et al. (2003) call this the 'loose mode of co-ordination in the EU'.
(2) Binding rules are necessary when the short-term preferences of different actors are inconsistent with their long-term preferences. Voluntary adherence to co-ordinated policy action does not ensure compliance over time, as the actors are autonomous and free to withdraw from the agreement in the future. By making binding commitments to an agreed long-term goal, dynamic consistency between multiple policy plans can be established. This poses another problem: will different autonomous actors honour their commitment to the policy rule over time? If the preferences of different actors are disjunct and do not converge over time, non-compliance is highly likely and a regime of hard sanctions is required in order to deter deviating behaviour. The regime of sanctions changes the pay-off matrix and therefore the incentives for strategic behaviour. Hard policy sanctions provide Pigouvian distorting incentives. The typical example is the Excessive Deficit Procedure, which has been further 'hardened' by the sanctions envisaged by the SGP. Explicit Treaty provisions establish the coercive mechanisms as legally binding obligations. I will return to the question of legality below.
(3) Between hard and loose modes of co-ordination stands the intermediary regime of soft rules such as the 'guided' co-ordination of the BEPG and the Luxembourg Process. If policy preferences between actors converge, they are more likely to act in a similar and consistent fashion given a specific set of circumstances. A policy rule is then necessary to prevent dynamic inconsistencies, but non-compliance by individual actors is less an issue and therefore requires only 'soft' sanctions. 1 Soft coercive mechanisms, such as 'naming and shaming', are not legally binding, although the exchange of information, collective learning and peer pressure do contribute to the convergence of preferences. Agreed policy documents, like the BEPG or the National Action Plans for Employment, formulate rules as guidelines or focal points for medium-term policies, while the implementation of policies is revalued on an annual basis, allowing a certain degree of discretion in policy-making.
(4) Finally, the community method of conferral delegates policy-making competencies to a unified agent such as the European Commission or the ECB. Policy decisions are no longer made by autonomous actors with responsibilities to different constituencies. Instead, the unified authority formulates and implements policies with respect to an enlarged constituency, internalizing all externalities by legally obliging other actors or governments. Conferral is the appropriate policy regime when national policy preferences are disjunct and discretionary decisions need to be taken, for otherwise anarchy and conflict dominate and prevent welfare improvement.
These different policy regimes are summarized in Figure 1. The two dimensions are preference consistency between actors and time consistency between actions. Disjunct preferences require 'hard' regimes of legally binding policy decisions in order to internalize the spillovers of externalities. If dynamic consistency can be achieved by formulating a general obligatory policy rule, then co-ordination of autonomous actors is possible, provided they stick to the rules. If, however, policy decisions need to be revalued frequently and actors have disjunct preferences, they may be better off in delegating policy- Figure 1 Policy co-ordination regimes making to a centralized authority acting as the agent of a larger constituency. Both regimes pose questions of democratic legitimacy. But, if heterogeneous preferences converge, owing to public deliberation, mutual persuasion and learning, then the simple exchange of information and less constraining forms of policy co-ordination may be sufficient to improve welfare. Voluntary cooperation is appropriate for specific discretionary issues, while a soft and flexible framework of policy rules would allow the integration of voluntary cooperation into dynamically consistent policy programmes.

Collective goods: linking externalities and co-ordination regimes
We may interpret the outcome of policy decisions as collective goods. The theory of collective action has shown that a regime's effectiveness to produce these collective goods depends largely on the nature of collective goods.
The classical definition of public goods is that (1) once produced they are available in the same amount to all the affected consumers (jointness of supply) and that (2) everyone must consume the same amount of the good, i.e. the supply of the good cannot be restricted to those who produced it (nonrivalness for consumption). However, as Figure 2 shows, the concept of collective goods is larger. If only condition (1) holds but not (2), the collective good is called a club good; in the opposite case it is a common property resource.
Collective goods, characterized by different externalities, require different forms of governance. However, the link is not a simple transposition from Figure 2 to Figure 1. The crucial distinction lies in the nature of expectations formed by autonomous actors regarding their pay-offs and the implications for interactive equilibria. Club goods in box III of Figure 2 are inclusive goods in the sense that (below a given capacity limit) an existing club may increase the total benefit from membership by admitting new members (Olson 1971). In this case the benefits of individual members are complementary to each other. The recent literature on network externalities has also observed the positive correlation between benefits and the number of group members (Katz and Shapiro 1985).
The interactions between agents determine the level of pay-offs and can therefore be related to collective goods. Cooper and John (1988) have provided a general framework that goes one step further to the level of actors' strategies. 2 They show that strategic complementarities arise if an increase in one player's strategy increases the optimal strategy of the other players. This can lead to multiple symmetric Nash equilibria, where mutual gains from a possible change in strategy may not be realized, because no individual player has an incentive to deviate from the initial equilibrium. Thus, strategic complementarities would prevent an optimal allocation of inclusive collective goods, unless a mechanism is in place to ensure that co-ordination failure is overcome. It can be shown that inefficiencies owing to strategic complementarities can be Pareto-improved by ensuring that all agents have equal access to information, allowing them to reap the full benefits offered by network externalities (see Benassi et al. 1994).
These models are useful for understanding policy co-ordination. In fact, the positive spillover of a given common policy on any individual jurisdiction would only take place if all individual governments were convinced that their marginal contribution to the common policy has a positive effect on their own constituency; otherwise they will abstain. In this case collective preferences between autonomous actors will converge, and this is exactly what the soft policy co-ordination regimes in boxes III and IV of Figure 1 are supposed to achieve: increased flow of information, exchange of best practices, peer pressure are all mechanisms to 'tip' a policy network away from an initial sub-optimal to a higher Pareto-dominated equilibrium. This logic has two implications: (1) Owing to strategic complementarities, voluntary policy co-ordination and adherence to guiding rules will be forthcoming if, and only if, soft institutional structures ensure the 'tipping'. (2) The group of decision-makers can be large, as network externalities increase strategic complementarities. Hence voluntary and soft policy co-ordination is the appropriate regime for the provision of inclusive collective goods. Enlarging the EU may increase the overall utility of European integration for all members.
However, the case is different for exclusive collective goods. Here policy coordination is hampered by strategic substitutabilities, which can be caused by two different forms of externalities. (1) Although consumption of inclusive collective goods is the same for all group members, their access (supply) is rival, meaning that the share of benefits for each group member falls as the number of participants increases (Collignon 2003). As a consequence, the feedback to the expectations of existing group members is negative. This creates strategic substitutabilities (the second cross derivative of the utility function is negative). The individual marginal benefit for a new or deviating member of the group is positive, but for existing or conforming members it is negative. Thus, there exists an incentive to go against the collective interest (such as the SGP) insofar as existing members' willingness to pay for collective goods will be reduced as the number of group members increases. Large groups will therefore become 'latent' (Olson 1971), i.e. the likelihood of a large group providing an exclusive collective good voluntarily will diminish. Thus, EU enlargement will destroy the benefits from European integration for exclusive collective goods, if left to voluntary policy co-ordination between member states. (2) Because benefits are the same for all group members, but access is rival, it is possible to externalize production costs to others and free-ride. For example, unco-ordinated borrowing in the financial markets by any public authority could cause externalities with respect to interest and exchange rates or inflation. However, the free-rider benefits from low interest rates 'produced' by other member states' collective fiscal restraint and this creates a generalized incentive for non-compliance. Therefore, hard forms of policy co-ordination or even, as in the case of monetary policy, delegation to a European authority are required to provide these collective goods efficiently.
The two points can interact. The zero-sum game quality attached to common resource goods provides an incentive to keep the group of collective decision-makers small, for otherwise one risks the non-provision of the collective good. 3 But if the group is enlarged, the provision of collective goods will be reduced and ultimately cease. Hence, intergovernmental voluntary policy co-ordination will not be sufficient to provide large groups of countries with an efficient or even optimal amount of collective goods. 4 A consistent approach to policy co-ordination for such goods requires 'hard' regimes of governance. Thus, boxes I and II in Figure 1 are governance regimes necessary for the provision of exclusive collective goods. However, it is also clear that the regime described in box II, where governments keep their autonomy, will become increasingly less effective as the EU enlarges to new members. Strategic substitutabilities will undermine adherence to and compliance with the rules however 'hard' they may be. This is the 'collective action problem' of European integration. It implies that governments cannot be the ultimate source of European political legitimacy. Because the problem becomes more acute after enlargement, a profound rethink of Europe's economic governance in a union of twenty-five or more member states is unavoidable unless half a century of integration is to be undone.

The efficiency of today's regime
How efficient is the EU's economic governance? We would expect that soft co-ordination regimes may be sufficient to provide inclusive collective goods. But even hard co-ordination regimes are no guarantee for the optimal provision of exclusive goods and then more conferral is needed. Policy processes have focused on providing inclusive collective goods by intergovernmental coordination that display, by definition, quasi-network externalities. Examples are 'Lisbon policies' for the Information Society, such as the deregulation of telecommunications, cyber space, digital TV, or creating a single sky. These network goods also have spillover effects on other sectors like human resources, agreeing on a Community patent or integrating a Europe-wide financial market. By following the benchmarks set out by the co-ordination process, member states increase their competitive advantages. Hence, there are strategic complementarities between their actions and it is not surprising that European integration has progressed in these areas in recent years.
However, these successes apply mainly to inclusive public goods with network externalities. For inclusive goods outside networks, progress is much slower, as witnessed by the liberalization of energy markets, completing capital market integration, or research policies. It could be argued that the monopolistic provision of national utilities, like electricity, is better classified as an exclusive collective good than a network externality. Little progress has been achieved on social cohesion, pension reform or a European framework for social protection that would facilitate cross-border labour mobility. Instead, converging in their preferences, national governments protect special interests relevant to their national constituencies, often preventing the implementation of previous agreements. 5 While soft co-ordination seems to work for inclusive goods, the policy regime of hard co-ordination is facing difficulties, not only with efficiency, but also with legitimacy. There is now a wide range of public policies in the EU creating exclusive collective goods. Fiscal federalism has established macroeconomic stabilization policy as an important subset of exclusive collective goods, which are dominated by strategic substitutabilities. They require policymaking at the centre, but in the EU that is not the case. The SGP is an example of the well-known problems of dynamic consistency. Much less discussed are issues of legitimacy related to collective preference change. Most analytic models assume given policy preferences and derive government actions from utility maximizing behaviour. Given the autonomy of national budget policies and the preponderate role of national parliaments, national policy preferences are exogenous and disjunct. Governments then bargain for solutions, which make the distribution of costs and benefits acceptable for them or for their constituency. However, neither the outcome of the bargain, nor the bargaining process itself, will change preferences within the national framework. Thus, EU preferences are heterogeneous and there is no mechanism to facilitate European-wide preference change through the collective deliberation of parliaments, citizens and voters. As a consequence European policies suffer from a 'democratic deficit' in the sense that citizens feel disenfranchised from policy decisions.
The co-ordination problem is reinforced by democratic preference change within national constituencies. For example, some recently elected governments see the SGP as an obstacle to the implementation of policies for which they have been elected. Yet, the purpose of the Pact was exactly to prevent discretionary changes in fiscal policy. But then the question arises: why have elections, if nothing can be changed? We are touching here the core of democracy.
There is also an alternative line of criticism. Fiscal policy should be more 'flexible' to counter insufficient demand in the business cycle. Yet, it is not clear that the SGP prevents this flexibility. In the early EMU years there was clearly a lack of commitment to balance structural deficits in several countries and pro-cyclical tax cuts deteriorated structural deficits. The 'sinning countries', Portugal, Germany and France, made least adjustments during boom times and are now calling for more discretion in fiscal policy. The theory of collective action tells us that if one were to increase 'flexibility' in national budget positions (not in the aggregate stance!), the most likely outcome would be more volatile monetary policy with less macroeconomic stability, because more discretion in national budgets would create an unco-ordinated aggregate stance. The ECB, acting as a Stackelberg follower, would have to adjust interest rates more often and this creates instability in financial markets with negative consequences for output. Hence, what is needed is flexibility in the coordinated aggregate budget position. The question then arises how rulebased co-ordination between autonomous fiscal authorities can provide policy discretion at the European level. Because stabilization policy is an exclusive good, collective action problems are likely to prevent this outcome. Therefore, the determination of the aggregate fiscal stance should be delegated to the European level. This leads us to a second fault-line, which is the democratic legitimacy of policy decisions at the European level.

The fatal flaw: lack of democracy
All theories of democracy presume that people who live together in a society need a process for arriving at binding decisions that take everybody's interests into account. One common justification for democracy allies the premise that people are generally the best judge of their own interests with the argument that equal citizenship rights are necessary to protect those interests (Gutmann 1993). The first premise refers to citizens' equal distribution of the power to make collective decisions (Warren 2001). Its frame of reference is the outcome of the decision process, as that is what affects citizens' interests. Different public goods may affect different groups of citizens, but collective European goods (i.e. policy outcomes) affect all citizens of the EU or at least Euroland. The democratic deficit arises at the European level because citizens are not participating equally in the deliberation of European policies and their power to make decisions is unequally distributed. Communication and deliberation on European policies is concentrated on national political competition and constituencies, the only arena where people can participate in policy-making. But as a consequence, political preferences for European collective goods converge to disjunct national rather than to European preferences. Furthermore, the ultimate power is with the Council over which European citizens have no equal rights. This is not only an argument about voting rights, but also about the impossibility of removing European decision-makers collectively.
National governments are accountable to their national constituencies. If these constituencies have national rather than European preferences, democratic choice through elections will ultimately have primacy over negotiated intergovernmental co-ordination. The conferral of budget policy to the EU level is therefore problematic. The Maastricht Treaty has left responsibility for fiscal policy at the national level and has therefore also confined deliberation of what is best to a level where the interests of all affected citizens are not taken into account. The hard co-ordination rule in the SGP -keep structural deficits in balance -is set for ever and applies to each member country, regardless of how preferences change. Inevitably this must lead to the de-legitimization of European policies.
The choice of an efficient policy mix among an infinite variety of consistent combinations of monetary and aggregate fiscal policy must be made at the EU level. An optimal policy mix should maximize European citizens' preferences, not those of governments. The democratic principle applied to delegation would require a European constituency.

The constitutional challenge
If the EU wants to sustain and improve its legitimacy and ultimately its power, it must function as a democratic society. Hence, it will need a mechanism to formulate policy preferences by European citizens at the European level. This is less of a problem for inclusive collective goods, where strategic complementarities provide the incentive for endogenous preference convergence and where the OMC between governments will yield results. But when exclusive collective goods demand at least some degree of discretion in policymaking, not only delegation to the EU level is necessary, but also the accountability of European policy-makers to a singly European constituency. For example, the definition of the aggregate fiscal policy stance is an exclusive good in Euroland, as it may affect inflation, interest and exchange rates. It needs to be delegated to the European level, for only there can it be determined coherently as the complement to monetary policy. However, the definition of the aggregate deficit lacks democratic legitimacy, unless it commands the consent of the majority of European citizens. Of course, the collectively preferred fiscal stance is largely the outcome of the policy debate on distributive outcomes; but legitimacy requires that all the distributive arguments are considered for the decision. Yet, when collective deliberation, communication and judgement only take place in national constituencies, the emerging consistency of preferences is national and not European. Hence, the binding force of the democratic process for collective decision-making does not extend to European public goods.
The formation of collective preferences across Europe requires democratic competition between rival élites for rival policy agendas and the involvement of ordinary citizens in European-wide policy deliberation (Hix 2003). Clearly, the control of the EU policy agenda by national governments does not allow democratic competition. What is at issue here is the nature of political cleavages: should European policy-making be dominated by national-culturalcommunication cleavages or by cleavages of political choice? With intergovernmental forms of governance, the dominant cleavage is national and this fact re-enforces the power and influence of political majorities in each member state at the expense of minorities. The EU's governance reflects, therefore, a cumulative cleavage. But cumulative cleavages also create legitimacy crisis, which could cause a collapse of the system (Dunleavy and O'Leary 1987). By contrast, establishing democratic structures for collective choice at the European level will reduce the impact of these traditional cleavages. This would also support European integration by encouraging trans-border political alliances below government level.
The active involvement of citizens in EU-wide policy deliberation will happen if, and only if, they become voters who can censure a European government accountable to its citizens. In the EU this is not the case. European policies are largely defined by the Council, but governments' accountability is to their national and not European constituencies. Political competition takes place within member states, not within Europe. As a consequence, national interests dominate collective European interests; the universal is dominated by the particular. The 'common concern' is substituted by the concern of national governments to be re-elected. Therefore, what is needed is effectively a govern-ment of Europe. This is not a statement of faith, but the obvious conclusion from the logic of collective action. Exclusive collective goods are not supplied optimally by large groups of countries on a voluntary basis. Conferral of decision-making to a European authority is necessary to improve welfare. A European government must reflect European-wide policy preferences, but it would also contribute to their emergence through being elected by and accountable to European citizens or their representatives in the European Parliament.
Some argue that fiscal policy-making should remain at the national level because national policy preferences are heterogeneous and centralization creates preference frustration and other inefficiencies. This argument for subsidiarity is wrong. If a European government is liable to its citizens, the political process of electing such a government creates structures of policy competition. European-wide deliberation will then reduce preference heterogeneity and policy dissent, for the competition for individual votes by parties and candidates requires them to explain their policies and engage the electorate. Before choosing, voters seek information. New information will change old preferences. An integrated polity across Europe will emerge as a result of establishing a European government. By contrast, the degree of dissent and policy frustration is structurally higher if policy-making remains the exclusive domain of natural governments. 6 This has implications for stabilization policies in EMU. The weakness of the SGP is not its short-term inflexibility, but its lack of democratic legitimacy in the long term. Certainly, hard budget constraints for fiscal authorities are necessary; they are only sustainable if they are backed by collective acceptance, expressed in the democratic choices of European citizens. The purely procedural argument that the SGP has been ratified by democratic governments is not a sufficient guarantee of its success or long-term legitimacy. It is necessary to define the macroeconomic policy stance at the European level, reflecting the democratic policy preferences of all citizens. (For practical implications, see Amato 2002;Collignon 2003.)

CONCLUSION
Europe's economic governance seems not only to be highly complex and of doubtful efficiency, but it is also unlikely to be sustainable in the long run. The OMC as created under the Lisbon Strategy will not optimally provide collective policy goods required to make Europe 'one of the most competitive economies' in the world. This method is only able to achieve results in the domain of inclusive collective goods that create strategic complementarities and common synergies. But as a result of setting up EMU, a large new class of collective goods exist that cannot be efficiently provided by voluntary cooperation between autonomous governments. They require not only policy co-ordination through hard rules to constrain deviating behaviour, but also full democratic legitimacy for their implementation. The EU runs the risk of being thrown into the dustbin of history by a revolution, which would have little to do with the noble principles of Jean Monnet. The solution is an EU with a truly democratic constitution, putting European citizens at the centre. Anything else risks the disintegration of Europe, given the increasing difficulties of voluntary co-ordination in an enlarged EU.