Debating the EU's Raison D'Être: On the Relation between Legitimacy and Justice

Since the eurozone, refugee and Brexit crises, it has become evident that the EU contributes to divergence among member states and inequality within them. It is also evident that the days of permissive consensus, integration by stealth, and trust in promises of peace and prosperity are long gone. Publics across the EU now wonder why the EU ought to garner their respect, and what role, if any, the EU should play in mitigating both cross&#8208; and intra&#8208;national socioeconomic inequalities. They wonder, that is, what gives the EU legitimacy beyond securing a growth dividend and protecting against another European war, and what criteria of justice we ought to use in assessing the distributional consequences of integration. This contribution articulates a conception of legitimacy that has received very little attention in the literature, namely telic legitimacy, distinguishes it from conventional criteria of &#8216;input&#8217; and &#8216;output&#8217; legitimacy, explains its usefulness in addressing the EU's crisis in confidence, and outlines its relation to principles of justice.


Introduction
In the wake of the eurozone, refugee and Brexit crises, it has become evident that the EU can reinforce not only growing economic divergence among member states but also within them. This has given rise to renewed calls for enhancing the EU's credentials as a 'Social Market Economy'. But how do we evaluate the EU's success or failure in this endeavour? Against this background, it is no longer sufficient to justify European integration in terms of the attainment of economic growth in individual countries, or the absence of armed conflict at the regional level. The unequal distributional outcomes of integration and interdependence have become all too apparent. Nor can we continue to take it for granted that deeper and broader integration is an end in itself. In an era when the ends the EU ought to promote can no longer be either assumed or simply deferredthe days of permissive consensus and integration by (technocratic) stealth are long gonewe need normative accounts of legitimacy, and, indeed, socioeconomic justice more than ever. Accounts of legitimacy guide us in our judgment about whether, and when, the EU ought to garner our respect. Principles of justice for the EU guide our assessment of the distributional consequences of integration. This contribution is an attempt to articulate a novel conception of legitimacy that has received very little attention in the literature, namely what I will call telic legitimacy, to explain its usefulness in assessing the EU, and to explain its relation to principles of justice.

I. Political Legitimacy and Coercion
Philosophers have traditionally focused on three broad concepts of political legitimacy. According to the first, an institutional system is legitimate just in case it has a moral permission to use coercive power to enforce its directives. According to the second, an institutional system is legitimate just in case it has a moral permission to use coercive power to enforce its directives, and subjects have a duty either to comply or, at the very least, not to interfere with enforcement simply because a directive has been issued. 1 According to the third, an institutional system is legitimate just in case subjects believe that it has a moral permission of one of the first two kinds (whether or not, that is, the institutional system actually has such a moral permission). 2 Notice that each concept focuses on the exercise of specifically coercive power, which is taken to be the hallmark of political authority. 3 As a result of this focus on coercion, the main task of the political philosopher is taken to be to explain how institutional systems can acquire the power to curtail our natural liberty in such a profound way. This is why any discussion of political legitimacy in either of the first two (normative) senses quickly resolves into a dispute with the anarchist.
The traditional emphasis on coercion and the obligation to obeywhich in turn flows from an abiding concern with a specific institutional system, namely the stateobscures a broader concept of political legitimacy. Consider, for example, that it makes sense to ask about the legitimacy of an institutional system (or aspect of an institutional system) even if it does not wield (or claim to yield) any coercive or enforcement power. 4 We can ask, for example, about the legitimacy of the UN General Assemblyand this we can do even though the General Assembly doesn't purport to rule, nor to authorize the exercise of coercion, nor to create any sanctions of any kind for failing to comply with the resolutions or declarations it makes. Once we open things up in this way, we can also ask about the legitimacy of an NGO, or the legitimacy of a treaty body whose task is to monitor compliance, but that otherwise lacks the ability to enforce its recommendations, or indeed the legitimacy of an institution such as Apple. This is an advantage for our argument with respect to the EU since we no longer need to worry whether and when the EU exercises any coercive power directly or only indirectly via the enforcement capacities of its member states, who retain ultimate control over the legitimate use of violence.
What broader concept of legitimacy might we appeal to in making sense of these possibilities? An authority is legitimate when its directives give us, as I will say, content-independent and exclusionary reasons for action. These reasons can be reasons not to interfere with the operation of the institution (for example, if we are not its addressees), or reasons to comply with it (if we are), or reasons to promote its ends (in ways that we might not otherwise have in its absence). 5 A directive gives rise to content-independent reasons when the reasons to comply, not to interfere, etc., do not depend on the nature or character of the action mandated by the directive, but on the mere fact that the directive was issued by the authority. Furthermore, the directive-dependent reasons are not just added to the directive-independent reasons we might have to do x, but exclude at least some countervailing reasons. For example, when the referee calls time, the very fact that she has determined that the game is over does not compete or simply add on to whatever reasons I have for continuing to play. The fact that the referee is an authority, and has called the game, excludes whatever reasons I had for continuing to play.
There are two aspects of this very general concept that are worth noting. First, the concept is broad enough to encompass non-coercive institutions: institutions can give us content-independent, exclusionary reasons by issuing directives even if they have no coercive power at all (think of the umpire in the game). They can thus be assessed in terms of whether their issuing of such directives actually gives us valid reasons to comply, not to interfere, or to promote their ends; they can be assessed, that is, in terms of their legitimacy. This is already a significant improvement over traditional accounts.
Second, in classical accounts of legitimacy, it is often unclear whether legitimacy assessments are meant to be binary or scalar. Is the legitimacy of an institution a yes or no matter? Or can an institution be more or less legitimate? On the view I will defend, an institution's degree of legitimacy is both a function of the breadth of the reasons it excludes within a given domain (relative to other similarly situated authorities) and the weight of the reasons generated (relative to those considerations that are not excluded). An institution, on my account, has less legitimacy to the extent that the range of considerations which it excludes in a given domain is narrower than some similarly situated alternative, and the extent to which the reasons it provides are weaker (relative to nonexcluded reasons). An assessment of the degree of an institution's legitimacy is therefore always comparative. 6 For example, a fire marshal's authority is more legitimate in this sense (its directives exclude a greater variety of reasons, and are more weighty) than the authority of a person with no training who simply takes over during a fire. There will be thus be a greater variety of reasons that could legitimately compete with the reasons generated by the person's issuing of a directive. The comparison can be to the same institution at a previous point in time. In this sense, an institution can lose legitimacy with respect to a domain of action due, for example, to changes in its structure or in more general circumstances. Within this framework, we can also preserve the judgment that some institutions are illegitimate. An institution is illegitimate when, relative to some domain, its issuing a directive doesn't give us any content-independent, exclusionary reason to comply with it, or not to interfere with its activities, or promote its ends.
To sum up, an institution's legitimacy is a function of the extent to which its directives give us, within some domain, content-independent, exclusionary reasons either to comply, not to interfere, or to otherwise promote its ends. The more an institution is legitimate, the more reason it gives us, in other words, to accord a kind of standing visà-vis our actions. Another way to put the point is to say that legitimate authorities deserve our respect. But having delineated a very general concept of legitimate authority, what do we say about the standards or criteria for deciding whether a political authority actually is legitimate? When does an institution generate valid exclusionary and content-independent reasons for action?

II. The Function of Legitimacy Assessments
To develop criteria and standards of legitimacy for a given institution, we begin by asking: What are assessments of legitimacy for? What social role or function are legitimacy assessments meant to play in political life? Answering this question correctly will help us, later on, to identify why assessments of legitimacy are so different in nature from assessments of what we might call institutional optimality (including optimality according to a conception of ideal justice). Allen Buchanan has argued that the function of legitimacy assessments is, fundamentally, to determine whether institutions are worthy of our reason-based support, where that support does not depend solely on the fear of coercion or on a perfect fit with the self-interest of those subject to it (Buchanan, 2018, p. 73). This makes sense: In order to coordinate action sufficiently to deliver essential goods and services, institutions must be capable of relying on attitudes of respect and support that are widespread and reliable. Relying solely on coercion would require a centralization and expansion of police power that is dangerous and undesirable. And relying solely on self-interest would be not only unreasonably limiting, given that institutions must often make decisions that are against our interests, but also lead to political settlements that are precarious and unstable, given the sheer variety and heterogeneity of individuals' purely self-regarding interests.
The function of legitimacy assessments is, in short, to provide reason-based, focal points for co-ordination in the presence of conflicting interests. Standards of legitimacy must therefore be (a) publicly justifiable and intelligible, (b) avowable in the presence of reasonable disagreement about their content, (c) specific enough to rationalize, organize and discipline the pursuit of feasible lower-level policies, plans and courses of action across time and across the institution. 7 The legitimacy theorist's task is therefore a practical one: Keeping in view the constraints and roles just mentioned, what standards and criteria could satisfy these three desiderata?
This way of conceiving of the function of legitimacy assessments gives us another advantage over classical accounts: it allows us to account for variation in the standards appropriate for different institutions. If the point of legitimacy assessments is to form a set of publicly shareable criteria and standards to evaluate whether a particular institution deserves our reason-based support in the presence of conflicting interests, then those standards and criteria ought to be sensitive to the kind of institution it is. Different standards and criteria will be appropriate, for example, for tennis clubs, multinational corporations, states, international organizations such as the WTO, and the EU. Such differences in criteria and standards of assessment are generated by the different relations in which such institutions stand with respect to their addressees (and with respect to other institutions), the scope of their competence, and the nature of the goods they seek to secure. This kind of variation is entirely overlooked by theories of legitimacy that only focus on one kind of institution, namely the state. But the variation allowed presents us with a problem: How do we get from the general concept of legitimacy to particular conceptions of legitimacy for specific institutions?

III. Standards and Criteria of Legitimacy: Introducing Telic Legitimacy
A natural place to start in answering this question is to identify general standards and criteria that any more particular conception of legitimacy ought to specify. Buchanan, for example, lists five: first, the institution should be reasonably efficient and effective (and certainly better than the non-institutional alternative); second, there must not be too large a discrepancy between its professed goals and what it actually does; third, it must be minimally morally acceptable (that is, it must not violate human rights); fourth, it must have acquired authority through some morally acceptable process or procedure; fifth, its procedures should be sufficiently fairor at least fair enough that it can be seen to work in the name of all those subjected to it (Buchanan, 2018). What counts as a relevant standard of 'fairness' or criterion of 'efficiency and effectiveness' will then vary by institution.
Buchanan's proposals strike me as uncontroversial, and reflect different aspects of what the literature on the EU often refers to as 'input' and 'output' legitimacy, where standards (1)-(3) spell out aspects of 'output' legitimacy (insofar as they specify requirements that apply to the decisional outputs of an institution), and (4)-(5) aspects of 'input' legitimacy (insofar as they specify requirements that apply to the decision procedures leading to the outputs). 8 In this section, I want to introduce a new general standard that is, I believe, at the crux of recent debates on the legitimacy of the EU yet rarely recognized as such. 9 It is also a standard that has received no attention in the theoretical literature on legitimacy, and that, as I shall seek to argue, deserves further attention.
This standard of legitimacy is, I want to claim, presupposed by traditional accounts of 'input' and 'output' legitimacy. I call it telic legitimacy. Take input legitimacy first. Suppose we are trying to determine what standards of procedural fairness, accountability, and responsiveness apply to an organization or institutionbut leave it open what particular organization or institution is in question. As we have seen, there is no all-purpose set of such standards. We first need to know something about the nature and character of the institution. Different standards will apply depending on whether, say, the institution is a central bank, a national legislature, the WTO, the EU, a university or a chess club. This is because standards of input legitimacy vary according to our account of the ends and purposes of an institution, and our assessment of its place in wider systems of institutions. But, if this is right, a full account of the institution's legitimacy will also need to assess whether the ends or purposes of the institutionwhat the institution is there to doare 8 The locus classicus for this view is Scharpf, 1999. It is unclear, however, what concept of legitimacy Scharpf himself has in mind. At points in the book, he seems to have a primarily empirical notion of legitimacy; at others, a more normative one. Andrew Moravcsik and I discuss this ambiguity in more detail in Sangiovanni and Moravcsik, 2002, pp. 122-148. 9 There is a large literature, of course, on the way the CJEU uses 'teleological interpretation' in its legal reasoning (most famously in its case on 'direct effect', namely Van Gend & Loos) according to which it construes the meaning of primary and secondary legislation in terms not only of the purposes of the narrow piece of legislation in question but also, and more controversially, in light of the general teloi of European integration itself. See, for example, Maduro, 2007, pp. 137-152;Lindseth, 2010 , pp. 701-712. What I say about telic legitimacy should not be understood as a defence of the Court's approach, or of teleogical reasoning in law in general. Rather, it is an attempt to say that structurally similar reasoning should underpin our understanding, as citizens, of the EU's legitimacy as such. I thank Anna Gerbrandy for helpful discussion. themselves legitimate. A criminal organization that is procedurally fair, effective and efficient, for example, will not be legitimate because its ends involve wronging others.
It might be thought that a criterion like Buchanan's (3)the idea that an institution or organization must be minimally morally acceptableis all that is needed to take into account the relevance for legitimacy judgments of the ends an institution or organization serves. But this, I want to argue, is false. Notice that (3) is entirely negative: it says that as long as an institution or organization does not violate minimal universal moral standards (such as human rights)and as long as it is effective, efficient, non-corrupt, and procedurally fair, accountable, and responsiveit is legitimate. On this view, there are no further, positive standards on which to evaluate the legitimacy of an institution or organization.
To see why this is too narrow a view of legitimacy, consider two examples where (1)-(5) are satisfied and yet where the question of the institution's legitimacy remains open. First, imagine that you are a professor at a university that decides to invest in areas that promise direct economic benefits for society at large, and to divest from areas that are unlikely to have any such impact. The university administrators argue that the point and purpose of a university is to function as an engine of national economic growth. 10 To justify taking money from the public purse, they ought to show how they can contribute to it, too. Let us further imagine that they have the means to pursue these objectives both efficiently and effectively, and that the decision violates no human rights (or other minimal moral criteria) or any criteria of procedural fairness.
Or imagine you are an employed activist at Amnesty International, but that Amnesty's objectives have recently changed radically: the Board of Amnesty decides to stop pursuing any of its other international efforts for the foreseeable future (including monitoring and reporting efforts), and instead to dedicate all its resources to targeting the Israeli government, with an explicit aim to topple the current regime. Let us further suppose, as in the university case, that the decision violates no human rights, is taken in accordance with Amnesty's usual procedures, which are procedurally fair and accountable, and would be effective and efficient in pursuing its new goal. As an activist employed by Amnesty International or as a university lecturer, we might still wonder: Does the change of course merit respect? Do we still have content-independent, exclusionary reasons to conform our will, against our interests and against our preferences, to the institution? I submit that despite the procedural and instrumental merits of both organizations (both meet all of Buchanan's criteria [1]-[5]), the answers to the questions I just posed remain open: criteria (1)-(5) might be necessary for legitimacy but they are not sufficient. But then what should their answers depend on? We need to take into account whether Amnesty or the university are pursuing the right sort of objectiveswhether they are pursuing the right kind of aims, ideals and valuesgiven their place in a wider system of institutions and given the role that legitimacy assessments are meant to play. We may claim that what gives Amnesty its legitimacy is its neutrality, its tradition of protecting human beings independently of particular political or ideological struggles, and its attempt to be 'above politics'. This is what distinguishes Amnesty in the universe of NGOs and other inter-and trans-national organizations, and which explains why the organization merits our respect. Pursuing its new ends, we may further believe, will be self-undermining, further polarize politics in the Middle East, and perhaps even lead to the erosion of human rights not just in the Middle East but elsewhere as well. We may, therefore, come (rightly) to feel that we have therefore lost any content-independent, exclusionary reasons to conform our will to its new ends and ethos. While we have complied in the past with many decisions with which we disagreed, this was always because we felt that the overarching, constituent goals of Amnesty were worthy of support, and they were worth pursuing even in the presence of disagreement (including our disagreement). And we might feel the same way, mutatis mutandis, about the university.
The standards employed to make these judgments regarding the ends of an organization or institution are standards of telic legitimacy. 11 These standards provide a positive counterpart to the minimal moral criteria mentioned by Buchanan. A conception of telic legitimacy for a specific institution or organization, then, aims to provide an account of such ends and the (feasible) lower-level policies, plans and courses of action required to realize them.
Before moving on, it is important to emphasize that such ends do not, however, require the optimal functioning of institutions. To insist that political authorities be optimal in order to garner our respect would be self-defeating: it would, in the long run, produce outcomes that were worse for everyone, including worse from the point of view of the optimal standards we favour. As Buchanan writes, the whole point of the social practice of making legitimacy judgments is to achieve a compromise between institutional optimality, on the one hand, and the need to secure the social support that an institution must have if it is to provide its distinctive benefits without excessive costs, on the other (Buchanan, 2013, p. 180).
An account of telic ends must, as I have said, defend standards that are not only publicly avowable, justifiable, and specific enough to rationalize feasible lower-level courses of action but also leave space for reasonable disagreement over their content. An analogy is useful: an account of telic legitimacy provides the overall target but does not require that one hit the bull's eye. The closer we are to the bull's eye, the more legitimate an institution will be, but an institution only becomes illegitimate when we have missed the target altogether.
These differences between optimality and legitimacy assessments should be obvious to anyone who has ever belonged to a social or political movement. The movement (and the organizations that give it voice, structure and coherence) will often take positions and decisions that one thinks are wrong-headed. But given one's support overall for the causewhatever it isit would be unreasonable to withdraw our respect overall for the organization and its members every time it or they didn't do as we would like. If we value the final ends that the movement serves, we should be prepared to accept defeat, for example, in a vote or decision, without taking this as sufficient reason to abandon the movement. It would be unreasonable, in part, precisely because such an intransigent attitude, if generally adopted, would undermine the ability of the movement to achieve the broader aims for which it was founded, and which we ultimately endorse. Notice, however, that in cases like these we take for granted some conception of the ends and cause that the movement is there to serve in the first place; as we have just seen, should the movement stray enough from those overall aims and ends, then we would be right to withdraw our reasonbased respect for the movement.
One might object: But shouldn't such telic ends be determined, not by systematic normative reflection but by procedurally fair, accountable and responsive institutionsfor example, democratic institutions? If this objection were correct, then, one might continue, we shouldn't really worry about telic legitimacy, which should be left to the public to decide. If what I have said is right, however, it is the other way around: knowing what standards of fairness, accountability and responsiveness applyincluding what kinds of [democratic] institutions are required for input legitimacyrequires a prior account of an institution's telic ends. Different kinds of organizationcorporations, states, international organizations, NGOsrequire different standards of input legitimacy. While of course lower-level ends, policies and outcomes ought to be decided via, for example, inclusive and procedurally fair processes, and while of course telic ends ought always to remain open to debate and discussion, the very existence of a fair, accountable, responsive institution always presupposes some account of why and how it is worth coming together to debate and decide in the first place; similarly, the appropriate standards for responsiveness, accountability and fairness themselves depend, as we have seen, on the point and purpose of the institution or organization in question.
There is also a deeper incoherence in saying that the 'people' ought to decide what telic ends to pursue and systematic normative reflection on those ends somehow preempts democratic deliberation. 12 To see this, consider the EU. Imagine you are one of those peoplean EU citizenwondering for yourself what telic ends the EU ought to pursue. If you, along with every other EU citizen, answered 'let the people decide!' then no-one would decide anything, since everyone would be waiting for the rest to decide for them. An account of telic legitimacy, in sum, is meant to address publics at the point at which they are trying to make sense, for themselves, what the point and purpose of the EU ought to be.

IV. Telic Legitimacy and the EU
What ends ought the EU to pursue in order for it to be telically legitimate? We need to explain what the EU is for. Why is the EU worth preserving? Why not simply dismantle it? 13 Of course, the finalité politique of the EU has been debated for decades (most importantly during the debates on the ill-fated Constitutional Treaty), but these debates have been almost exclusively focused on the final form the institutions of the EU would take would the EU take on the form of a fully-fledged federation? Would there be a Philadelphian or Madisonian moment in which the EU would be transformed into something much closer to the US or Germany than to a looser confederation? In this tradition, the final ends the EU ought to serve were either taken as uncontroversial (peace and prosperity) or deferred until the endpoint was reached. In the meantime, it was thought, the EU could be legitimated entirely in terms of more technocratic ends, such as efficiency and stability. But now the permissive consensus that allowed such a thinly instrumental 12 Habermas, I believe, makes just this mistake in his criticism of Rawls in Habermas, 1995, pp. 109-132. 13 On the importance of articulating such teloi explicitly, see Morgan, 2004 ;Weiler, 2011. account of output legitimacy has ended. It is for this reason that, I have been arguing, we need to think more explicitly about telic legitimacy.
As Gary Marks writes, 'The European Union exists chiefly to gain the benefits of scale in providing public goods' (Marks, 2012). But which goods? The main collective goods that the EU aims to provide include a stable and predictable legal system (which forms the background for all other goods provided by the EU), an efficiency-and productivityenhancing single market (comprising inter alia a customs union, competition law, elimination of tariff and non-tariff barriers, and free movement of persons), a single currency for eurozone members (generating lower risk premiums for soft currency states, exchange rate stability, and lower transaction costs), a mechanism for being rule-makers rather than ruletakers in international bargains regarding, for example, trade, and regional and domestic stabilization both internally among members 14 and on the EU's periphery (the latter achieved in large part via the process and promise of accession) (Alesina et al., 2005;Grabbe, 2006;Schimmelfennig and Sedelmeier, 2005). Especially in view of the constitutional asymmetry between negative and positive integration (Scharpf, 1999), the predominant aims of the EU have been therefore mostly market-creating and market-maintaining.
But why and under what conditions are these ends worth pursuing? Why and under what conditions, that is, does the pursuit of these ends give EU citizens exclusionary reasons to support and maintain EU institutions? Note that 'prosperity for all' is an unconvincing answer, all the more so in light of the erosion of the permissive consensus. The forms of integration that have underpinned the collective goods mentioned above have distributional consequences both within and across member states. Some states will gain more than others from integration; other states, even though they might still benefit overall, will be harmed in specific issue areas. Think here of the inequality between Germany and Greece in the wake of the debt crisis, where German GDP per capita has grown by 7 per cent and Greece's has fallen by 20 per cent, or the differential impact of the refugee crisis. Similarly, some regions and indeed individuals will gain more than others from integration; others will be harmed. Think here of differential opportunities for exercising free movement rights across member states (Bruzelius et al., 2017), or the fact that European integration has increased intra-national regional inequalities even as it has, on average, diminished cross-national ones, or the implications for citizens of bail-in rules in the new Banking and Recovery Resolution Directive. 15 Given the increasing politicization of the EU's role in promoting markets but not fairness, it is no surprise that the Lisbon Treaty mentioned 'solidarity' over 20 times, and Article 3 TFEU declares, for the first time, a commitment to 'a highly competitive Social Market Economy'. With the eurozone, Brexit and refugee crises, the following question has become even more pressing since 2009: Yes, a Social Market Economy, but what kind of Social Market Economy? 16 In classical Polanyian terms, we need to know something more about the kind of society (or rather societies) a truly European market ought to be embedded in.
14 But see Müller, 2013, and the recent rule of law crises in Poland and Hungary for some of the difficulties in this area. 15 See http://bruegel.org/2016/11/income-inequality-has-been-falling-in-the-eu/; Beckfield, 2006, pp. 964-985;Dunford and Smith, 2000, pp. 169-195;Dunford and Smith, 1998. 16 For the notion of a Social Market Economy (including its historic connection to German ordoliberalism), see the Introduction to this Special Issue, and Joerges and Rödl, 2004. We need, in sum, an account of telic legitimacy for the Social Market Economy that goes beyond vague appeals to general prosperity and that takes into account the distributional consequences of integration in an era of constructive dissensus (Hooghe and Marks, 2009). How might one go about doing so? Here I can only provide a sketch designed to show the relevance of the desiderata mentioned above and, as we will see, the role of systematic normative reflection (including on justice) in accounts of telic legitimacy. We can contrast four general models for how to conceive of the ends that ought to be served by the idea of Europe as a Social Market Economy: 1. The EUespecially given EMUhas gone too far; it is illegitimate; the costs mentioned above are unacceptable. There are three main versions of this thesis. According to the first, the EU ought to be rolled back because of its corrosive effects on domestic welfare states (see, for example, Scharpf, 2015) and, on the second, because of its corrosive effects on the possibility of truly egalitarian policy (see, for example, Streeck, 2014;Tuck, 2016). According to the third, it ought to be rolled back because the emphasis on harmonization (including vain attempts to harmonize social policy) makes Europe less competitive, less productive, and less democratic; the EU should foster, instead, 'cooperative [regulatory] competition' at the inter-and supra-national level and leave social policy to democratic, and hence domestic, publics (Majone, 2014). 2. The EU is in an unstable halfway house. It ought to make a decisive move in the direction of federalization including in fiscal and social policy (for example, by granting the EU relevant and autonomous tax and spending powers). Were it to merely muddle through it would lose whatever legitimacy it currently has (see, for example, Habermas, 2015). 3. There is a stable middle ground. On one, more intergovernmental version, we should conceive of the EU as a system of co-operation among member states; a European Social Market Economy should therefore serve in the first instance to consolidate, stabilize and extend the problem-solving capacities of its member states (which may require, for example, forms of risk-sharing, risk-reduction, joint investment, and conditionality within EMU) (see, for example, Bénassy-Quéré et al., 2018;Sandbu, 2017;Stiglitz, 2016). If it doesn't, it should lose our reason-based support. On this view, the transnational dimensions of the EU (including, most importantly, free movement of persons and cross-national transfers) should be justified primarily in terms of their contributions to national-level productivity, solidarity and democracy rather than as ideals in themselves (see, for example, Moravcsik, 1998 andmore recently, Moravcsik, 2012). 4. As in (3), there is a stable middle ground. But according to this, less intergovernmental version, we should conceive of the EU both as a system of co-operation among member states and among EU citizens as such. On this version, while we can conceive of the EU as primarily a system of co-operation among member states, it is not exclusively. Transnational forms of co-operation (including cross-national transfers and the free movement of persons) have independent justificationindependent, that is, from justification in terms of the problem-solving capacities of member states one-by-one. 17 Were they to be rolled back, the EU would lose legitimacy (see, for example, Bellamy, 2017;Cheneval and Schimmelfennig, 2013;Nicolaïdis, 2013).
How should we adjudicate among these as conceptions of the EU's telic legitimacy? Recall the three desiderata that an account of telic legitimacy must meet to function as a focal point for co-ordination: an account of the ends of integration ought to be (a) publicly intelligible and justifiable, (b) avowable in the presence of reasonable disagreement about their content, and (c) specific enough to rationalize, organize and discipline the pursuit of feasible lower-level policies, plans and courses of action.
One might think that the most obvious way of adjudicating among our alternative models is to appeal to their empirical merits. For example, is it really true, as an empirical matter, that there is no alternative to federalization or to rolling back the EU? This would avoid the need for any normative judgments. But notice the difficulties inherent in any such empirical assessment. While it might be possible to give an account of the most likely tradeoffs one would face in adopting each model, it will be well-nigh impossible to exclude any as impossible on the basis of a merely empirical assessment. At the end of any such empirical assessment, we would still be faced with a choice: Which tradeoffs (taking into account their uncertainty) are the most appealing? For this further decision, we will need to assess what ends integration ought to serve, and so its telic legitimacy.
One might wonder whether one could exclude some models on the basis of their infeasibility (so desideratum (3)). But here too there are problems. First, judgments of feasibility are very difficult to estimate reliably. To see this, consider that feasibility is best understood as an empirical property of a course of action (such as a policy or policy outcome) consisting in the likelihood of realizing the course of action in question against some fixed background (see, for example, Gilabert and Lawford-Smith, 2012). Given the unreliability in general of forecasting models in the social sciences, and given how contrary-to-fact judgments of feasibility must be to be useful, it is likely that we will only be able to exclude only more far-out proposals. 18 Second, it will be very unclear what one should allow to vary in any such judgment (background economic conditions? which ones?) and what to hold constant (public opinion?) over what time frame (5, 10, 50, years?). Settling on a consistent set of such conditions across modelsthough it will not undermine coarse-grained comparisonswill vitiate any attempt to make convincingly fine-grained distinctions.
If this is right, any adjudication between the models will need to provide an account of their public justifiability and avowability in the presence of reasonable disagreement. And here too we might think there is a more empirical route: a model is publicly justifiable when it is most likelyalong with its profile of tradeoffsto be accepted by democratic publics across Europe. As a version of the 'let the people decide!' argument mentioned above, this is incoherent. We want to know, as a member of one of those European publics, why and under what conditions we have exclusionary reasons to comply and support the EU. To this question, as we have seen, it is incoherent to respond: You have reason to comply and support the EU if everyone else does. Public justifiability and avowability in 18 Majone, for example, rejects welfare federalism at the EU level as infeasible in Majone, 2014 . This might be plausible. But his primary argument, however, seems to be that monetary union can only ever undermine productivity and competitiveness across the EU unless there is either a full-scale transfer of powers to the federal level or EMU is abandoned (on this, see also Scharpf, 2015, pp. 384-405;Streeck, 2014). Given the disagreement about whether, in fact, there really is no other alternative, this kind of 'purely empirical' argument looks hasty. At most, there are tradeoffs which need to be evaluated, as I am arguing, on other terms. See, for example, Bénassy-Quéré et al., 2018;Sandbu, 2017 this context mean something else, namely providing a normative justification of the principles, values and idealsin short, the telic endsthat underpin commitment to each model mentioned above. 19 Given the distributional consequences and costs of EU integration, at the centre of any such account must be an explanation for why the particular model favoured serves to realize a fair distribution of the costs and consequences of integration. An account of the telic legitimacy of the Social Market Economy must have at its core, that is, an account of social justice that can help us in evaluating tradeoffs. Take the first model, which claims that integration has gone too far because it undermines the welfare state or prevents the realization of truly egalitarian policies or stifles productivity. For these arguments to be convincing, we need to know why, in the first case, we should keep the welfare state in its current form, what, in the second case, a truly egalitarian policy looks like, and, in the third case, why productivity ought to take priority to other values where there are tradeoffs. Considering that the model also requires rolling back integration, we also need to know whether the distributional consequences on other member states and their citizens as a result of doing so are acceptable from a normative perspective. Similarly, given the unlikelihood of being able to show that there really is no alternative, a defence of the federalization model is most convincing if based on a comparative normative judgment: would, for example, rolling back integration or stopping short of federalization in fiscal and social policy violate claims to justice that, say, the worst-off citizens in the EU have against the better off? Do member states and their citizens have obligations of, say, solidarity to support 'ever closer union'? If so, why? If not, why not? And, with respect to the third, 'in-between' model, the claim for a significantly autonomous transnational dimension requires an appeal to its value as against a more intergovernmental version. Does an autonomous transnational dimension, for example, rest on claims in justice that EU citizens possess to free movement? Might it rest on some other value distinct from justice? Are there reasons of justice for more 'risk-sharing' within EMU (including, for example, the creation of a risk-free EU asset, in which case: would Germany have a complaint in justice against the creation of such an asset?)? Are there reasons of justicerather than merely prudenceto be more (less?) vigilant regarding the possibility of moral hazard? In domestic politics, policy choices are always made in the shadow of concerns with social justice and fairness. It is surprising that, at the EU level, these kinds of questions have only rarely been asked in a systematic way (though they are gaining ground in public debates). This is not the place, of course, to provide a fully-fledged theory of social justice for the EU, which can then be used as part of a defence one of the four models (or some further combination). 20 It is enough for our purposes if I have shown how providing one is essential not just for its own sake but also for evaluating a crucial aspect of the EU's legitimacy, namely its telic legitimacy.

Conclusion
In closing, I want to clarify how we ought to conceive of the relationship between justice and telic legitimacy. I have argued that, when we ask ourselves whether we have exclusionary reasons to support and comply with EU institutions and policies, we ought to ask not just whether the EU is input legitimate, effective and efficient; we should also consider what ends the EU ought to pursue. Once we do so, we will inevitably come face-to-face with the fact that the EU has significant distributional consequences across individuals, regions and member states. Any account of telic legitimacy will therefore also need to take into account whether those distributional consequences are distributed fairly, and an account of such fairness will require an account of social justice. It might seem, therefore, that I have collapsed telic legitimacy and justice: telic legitimacy is justice by another name. This is a mistake for two reasons.
First, while I have argued that at the heart of an account of telic legitimacy for the EU will be an account of social justice, I don't believe this will be true of every institution (indeed, it may even be less important for other aspects of the EU's telic legitimacy, say, in the area of security). It will be less true, for example, for, say, a tennis club's telic legitimacy. I have argued instead that the EU's special circumstancesincreasingly evident since at least the early 1990smake an account of social justice unavoidable. Furthermore, it is important to remember that social justice is not the only end constituting the EU's telic legitimacy. As we have seen above, the EU's point and purpose is ultimately to provide a set of public goods (including sustainable growth via a currency union and single market, security, environmental protection, and so on); a full account of telic legitimacy would provide further detail regarding each one.
Second, assessments of an institution's social justice and assessments of its telic legitimacy have different roles in our moral reflection. An account of social justice provides an all-things-considered judgment regarding whether people's moral claims to distributive shares and social standing have been met. It has no immediate implications for whether institutions and policies give us exclusionary reasons to comply and support them. For that further questionthe question of legitimacywe also need to consider whether the ends, policies and courses of action currently pursued by the institution are avowable in the presence of reasonable disagreement. As I have said above, just because an institution does not hit the bull's eye (such as an ideal of social justice or, indeed, an ideal of democracy) does not mean that we lack exclusionary reasons to support and comply with it. To be sure, at some point, if the institution strays too farif it, say, actively promotes severe social injusticethen it ought to lose our respect.
But who should decide when it has gone too far? Each person, one-by-one, will need to decide for themselves whether this is the case or notkeeping in mind the importance of institutions as focal points for co-ordination, and the unreasonableness of insisting on our own point of view when others disagree. But there will be many cases in which we can say: 'the institutions are unjust or otherwise misguided, but, while we work to change them, they deserve our reason-based support'. The point I have made is that, for a judgment of this last kind to be intelligible, we first need to knowto make a case forwhat ideals, values and principles the institution should aim for in the first place, among which, in the EU case, will be an account of social justice. These provide the target and the bull's eye, without which we have nothing to aim at. We cannot defer this question to 'the public'or to some further proceduresince we are ourselves members of that public, engaged in debates on what the point and purpose of the EU ought to be, and hence, inevitably, what principles of social justice ought to guide our judgment.