Introduction
In his two Treatises on Government, Locke famously criticizes Hobbes for supposing that individuals should willingly give up the inconveniencies of an unruled condition (the state of nature) for the rule of an unconstrained sovereign power.
This is to think that men are so foolish that they take care to avoid what mischiefs can be done them by polecats and foxes, but are content, nay, think it safety, to be devoured by lions (Reference Locke1924 [1690]: 163).
Locke’s point is, of course, that individuals are unlikely to consent to a form of rule that could unduly expose them to harm. The restraint of power is, therefore, a condition for its legitimacy. At first sight, this seems unlikely to help us understand the problem of European Union (EU) legitimacy, save, perhaps, to beg the question of why a polity that so often appears constrained to the point of incapacity is seen to pose a problem of legitimacy at all (Barker, Reference Barker2003). Yet, as Philip Pettit has argued, we seriously underestimate the challenge of legitimation if we understand it as one of merely restraining power tout court, rather than one of identifying a form of political power that can be justified as ‘restrained yet capable’ (Reference Pettit1997) at the same time.
Now, that clearly is a challenge relevant to the EU. An understanding that legitimate power needs to be ‘restrained yet capable’ is already implicit in the tools that have been developed to study the EU’s own legitimacy. What else does the widespread practice of distinguishing the input and output components of EU legitimacy (Scharpf, Reference Scharpf1999: 2) signify other than the recognition that for EU institutions to be legitimate they must be capable of producing outputs the governed find useful; and yet constrained enough to satisfy the agreed procedural conditions for the rightful exercise of EU powers?
So why does power need to be both restrained and capable if it is to be legitimate? Pettit’s answer begins with the claim that individuals value non-arbitrariness in social systems and that they cannot secure that value by merely restraining institutions of government, since the latter are themselves needed to restrain arbitrariness originating in economic, social, and international relations (1997: 287–288). Thus, a need to control the arbitrary power of the monopolist to extract rents, the arbitrary power of the bad neighbour to impose negative externalities on others, the arbitrary judgement of the victim who turns out to be a ‘poor’, and disproportionately vengeful judge in his own case or, indeed, the arbitrary power of the bully in international relations. All these can and have been used to justify the empowerment of strong governing institutions. For Pettit, then, there is an internal relationship between restrained and capable forms of political power. Where either is justified, it is in relation to the same value, namely, that of minimising arbitrariness, which he defines as a condition where decisions depend only on the ‘arbitrarium’ (the pleasure or whim) of those making them, and not on the judgement, let alone the control, of those affected (Reference Pettit1997: 55).
Important though this answer is, it seems to me that the dependence of legitimacy on non-arbitrariness goes deeper. To see why, it helps to consider Jürgen Habermas’s definition of legitimacy as those ‘political obligations’ actors ‘put themselves under’ through the propositional logic of their own ‘moral claims’ (Habermas, Reference Habermas1996: 67). Although this definition is demanding – conceptually and practically – it has important advantages. By confining legitimacy to those obligations which the governed put themselves under of their own free will, it defines legitimacy in a way which avoids the circularity of assuming the justifiability of those very distributions of power or interest that are in need of justification. By limiting legitimacy to those commitments which follow from the moral claims of the governed themselves (see also Gaus, Reference Gaus2009), it allows for a key feature of liberal political orders. Since those orders are committed to the view that individuals should judge what is good and what is right for themselves, their citizens must be able to regard any coercion as no more than an enforcement of their own moral obligations on themselves (Rawls, Reference Rawls1993: 138). Two principles follow. The first is the principle which Habermas (following Rousseau and Kant) terms democratic self-legislation. Citizens, acting as political equals, should ultimately be able to control the making, revision, and administration of laws by which they are themselves coerced. The second is the principle that John Rawls terms political justice. Even democratic majorities cannot legitimately coerce obedience to particular views of the good drawn from a range of conflicting yet equally reasonable alternatives (1993: 138–139).
In this paper, then, I will use the twin principles of democratic self-legislation and political justice to evaluate the claim that the EU can be expected to be legitimate in so far as it approximates a ‘Coasian bargain’ between its member states. Far from this being a narrowly technical undertaking, it is pivotal to the study of European integration. Not only do influential contributions to the literature on integration make more or less explicit use of Coasian assumptions, but also they are right to do so from their own point of view. As I hope to show, Coasian bargaining theory (CBT) can go a long way to counter doubts that a body that is as ambitious as the contemporary EU in its institutional design and its policy scope can be indirectly legitimated by its member states. Indeed, I will argue here that if CBT fails, then so does indirect legitimacy.
However, while the Coase theorem is a social choice concept, the paper is a work of political philosophy. It aims to follow Amartya Sen’s advice that more use should be made of problems identified by social choice in analysing ‘moral philosophical issues’ (Reference Sen2002: 328). Thus, it proceeds as follows. The section ‘CBT and indirect legitimacy’ introduces CBT. The section ‘Social choice puzzles’ identifies difficulties in applying CBT to the EU from a social choice perspective. The sections ‘political justice’ and ‘democratic self-legislation’ consider the implications of those difficulties for democratic self-legislation and political justice. In effect, this research design extends Fritz Scharpf’s analysis of joint decision traps. It offers a somewhat different take on where consensus bargaining may produce sub-optimal outcomes; and, by clarifying the implications for democratic and political justice, it identifies the philosophical principles which justify Scharpf’s insight that bargaining problems may have delegitimating effects. Yet for reasons, I will set out in the conclusion, I am less confident than Scharpf that indirect legitimacy can be rescued from shortcomings in Coasian bargaining theory.
CBT and indirect legitimacy
Consider the following answer Giandomenico Majone gives to the question ‘which policies could be legitimately included in the agenda of a European federal state?’
Aside from foreign and security, the public agenda would mostly include efficiency-enhancing, market preserving policies … Unlike redistribution – zero-sum game – efficiency issues may be thought of as positive sum games where everyone can gain. Hence, efficiency-enhancing policies do not need a strong normative foundation: output legitimacy (accountability by results) is generally sufficient (2005: 191)
Thus, in Majone’s view, a Pareto-improving cooperation between member states would need little further justification. A glance at his references shows that Majone traces this understanding of justifiable collective choice back to Buchanan and Tullock’s Calculus of Consent (Reference Buchanan and Tullock1962) (see Majone Reference Majone2005: 55–56), which has in turn been described as a ‘political Coase theorem’ (Parisi, Reference Parisi2003). In what follows, I introduce CBT by considering the answer it suggests to the puzzle of how to secure a restrained yet capable form of collective choice. This will also allow me to distinguish between differences within CBT, which derive from its evolution from an economic theory about externalities (Coase, Reference Coase1960) to a theory that is also used in political science to analyse questions of institutional design (Buchanan and Tullock, Reference Buchanan and Tullock1962) and inter-state bargaining.
Ronald Coase (Reference Coase1960) demonstrated that as long as there are efficiency gains to be had from eliminating negative externalities, actors should, under conditions I set out later, be capable of identifying and realizing those gains, no matter how property rights are legally defined between themselves. Buchanan and Tullock argued that something similar could be said of political institutions. Institutions could employ any decision rule from dictatorship (one person decides) through simple majority voting to unanimous consent and still be capable of realizing all possible efficiency gains. Important contributions to the European integration literature likewise assume, first, that the main reason member states cooperate is to eliminate negative externalities (Moravcsik, Reference Moravcsik1993: 485); and, second, that bargaining between governments will usually be sufficient to identify sources of mutual gain. In this understanding, EU institutions are capable where they serve as an effective ‘contractual environment’ for bargaining between member states (Moravcsik, Reference Moravcsik1993: 508), notably by securing credible commitments (Moravcsik, Reference Moravcsik1998: 73; Majone, Reference Majone2001; Pollack, Reference Pollack2003), pooling access to sources of information and expertise (Majone, Reference Majone2001; Pollack, Reference Pollack2003) and dealing with problems of incomplete contracting (Garrett, Reference Garrett1995: 172).
Yet ‘political Coasians’ in the tradition of Buchanan and Tullock also see themselves as having a great deal to say about the ethics of restrained collective choice. By identifying conditions where Pareto improvements can be secured under any decision rule, they believe they also demonstrate that there need not be any objections of efficiency to preferring unanimity on the ethical grounds that only it can ensure the autonomy of individual actors. Where Pareto improvement is combined with consensus decision rules – where mutual gain is combined with mutual consent – it should be possible for individual actors to solve their collective action problems, while, at all times, deciding for what are their own moral reasons (1962: Ch. 18 and Appendix 1). Thus, in the case of a collaboration between states, we might define a Coasian scheme of cooperation as one where: (a) the scope of cooperation is limited to one of mutual gain between member states and (b) the process of cooperation remains sufficiently within the control of each participating state for it to be able to cooperate for its own reasons of value.
The powerful contribution which CBT can in principle make to the analysis of the EU and its legitimacy can be clarified further by answering two fairly obvious objections to applying CBT to the EU. One objection might be that the ‘facts’ of the EU simply do not fit Coasian assumptions. Some EU policies re-allocate values between member states and some EU decision rules impose unwanted obligations upon them. Yet, in answer to the point about re-allocative policies, ‘Coasians’ might reply that truly efficient ‘efficiency-seekers’ do not seek anything so crude as Pareto improvement in relation to single policies. Rather, they trade their vetoes, even if that means losing out here and there. As for majority voting, that can be seen, from a Coasian perspective, as little more than a closure device for the efficient and timely pursuit of objectives that are themselves supported by a consensus of member states (Moravcsik, Reference Moravcsik1993: 510). Not only, as Rousseau might have put it, does majority voting ‘presuppose unanimity, on one occasion at least’ (Reference Rousseau1973 [1762]: 173), but it is in practice used only sparsely (Mattila and Lane, Reference Mattila and Lane2001) and, it would seem, with regret in so far as the search for consensus often continues after member states have been outvoted. Comitology committees – which play the key executive function of agreeing implementing instructions – are often used to achieve real-time adjustments between common obligations and the preferences of individual governments (Sabel and Zeitlin, Reference Sabel and Zeitlin2007).
A rather different objection is that Coasian assumptions do not really address issues of legitimacy at all. In other words, they are a theory of mutual benefit, not of political obligation. To answer this objection, we need to understand what indirect legitimacy requires. Put simply, indirect legitimacy is the belief that the EU is legitimated via its member states on the principle that a body whose ‘authority is recognized and confirmed by the acts of other legitimate authorities’ is itself legitimate (Beetham and Lord, Reference Beetham and Lord1998: 11). The EU has at least one structural characteristic, which suggests that some element of indirect legitimacy remains essential to it. Its enforcement structures run through its member states. Fritz Scharpf puts the point thus:
If the function of legitimacy is to motivate compliance with undesirable obligations (his emphasis), what matters for the EU is the compliance of governments, parliaments, administrative agencies and courts within member states…Empirically, therefore, the EU is best understood as a government of governments, rather than a government of citizens (2007: 5).
Thus, Scharpf argues, enquiry into the legitimacy of the EU should focus ‘primarily’ on ‘normative arguments that could oblige governments to comply with undesired rules’ (Ibid. See also Scharpf, Reference Scharpf2009: 179–181). Rodney Barker likewise questions how far it is meaningful to ask whether the EU is legitimate in relation to individuals as opposed to its member states: ‘legitimacy is a concept which can usefully be applied to rule or challenges to rule. It cannot usefully be applied where rule is absent, hypothetical, or so indirect as to be invisible to the ruled’. He then puts his finger on what, in his opinion, it is about the EU which allows its legitimacy to be indirect. As he puts it, ‘the EU may govern’ but ‘it does not follow that it has subjects in the same way a state has’ (2003: 159–160). In other words, the EU does not, for the most part, require the obedience of individuals, since its laws are only enforced through the medium of national law. Thus understood, indirect legitimacy allows the EU to seek solutions to collective action problems which elude single states while relying on those same member states for the most state-like of their characteristics: their pre-acknowledged monopoly of legitimate coercive power.
Yet it is a mistake to view indirect legitimacy as a soft option. The consent and compliance of states can only be a necessary and not a sufficient condition for indirect legitimacy. The principle of democratic self-legislation plainly also requires that individual citizens should retain ultimate control over the making and administration of laws by which they are bound. Moreover, given a free and critical media, and temptations on governments to pose as the reluctant enforcers of laws made elsewhere, it cannot be assumed that the origins of EU laws in EU institutions can be hidden from national publics. Thus, as a matter of both principle and perception, attempts to secure legitimacy indirectly have to be justifiable for what they really are, namely, as a lending out – on terms individual citizens can accept – of national enforcement structures to laws made according to the procedures of the EU. Although, then, indirect legitimacy is self-evidently an attempt to derive the legitimacy of the EU from that of its member states, it also has to be capable of justifying a form of authority exercised by those states that is itself profoundly changed by their membership of the EU (Eriksen and Fossum, Reference Eriksen and Fossum2007: 14).
Thus, in the remainder of the article, I will make the following assumptions about the objects and subjects of indirect legitimation. The object of indirect legitimation – the entity that is in need of justification – is, as it were, ‘the state within the EU’: the member state, rather than the nation state; the state that obliges itself through its commitments to the EU to act as the enforcer for EU law in exchange for decision rights in the making of that law. However, the subjects of indirect legitimation – those to whom justification is owed – are the same as with any order governed by liberal democratic principles, namely, individual citizens, all of whom are equally entitled to a justification for why they should be bound by their own moral obligations to comply with coercively enforced law (above p. 2; Forst, Reference Forst2007).
Put another way, there are two conditions for indirect legitimacy:
(a) the governments of member states feel morally obliged to enforce commitments entered into at EU level; and
(b) citizens feel that their moral obligations to comply with national law extend to instances where it is used to enforce European law.
Now, it seems to me that CBT contributes to indirect legitimacy by identifying how conditions (a) and (b) can be expected to align and stay aligned. Here, the key point is not just that Pareto improvement implies member governments will have no obvious incentive to defect from their obligations. Of even greater importance from a point of view of legitimacy is the idea that the ethical neutrality of the Pareto criterion should also allow each member state the opportunity to cooperate on terms that establish its own legitimacy vis-à-vis its own public. Indeed, by making it possible to confine ethical choices, and any solidarism needed to sustain them, to domestic arenas, CBT also reduces the risk that cooperation at the European level will become domestically contentious in ways that need to drive a wedge between obligations to comply with EU law and the brute fact that governments depend on their domestic politics for electoral survival. I take it that Andrew Moravcsik has something along these lines in mind when he remarks:
There is an undeniable normative attraction to a system that preserves national democratic politics for those issues most salient in the minds of citizens, but delegates to more indirect democratic forms those issues that are of less concern, or on which there is an administrative or legal consensus (Reference Moravcsik2005: 6).
In answer, then, to the objection from which this discussion departed, we do not need to commit the category error of confusing a theory of bargaining with one of legitimacy in order to appreciate the importance of CBT to indirect legitimacy. The suggestion here is not that CBT confers indirect legitimacy. It is merely that it is one way of aligning the two conditions – set out in (a) and (b) above – which do confer it. Whether it is unique in that regard is a question for the conclusion. In the meantime, we need to consider problems in their application, both from a social choice perspective and in relation to legitimating values of political justice and democratic self-legislation.
Social choice puzzles
The last section showed that the structure of political obligation implied by indirect legitimacy is likely to work better where it can be assumed that the EU can operate as a Coasian scheme of cooperation. Yet there are well-known problems both with the Pareto criterion and with the Coasian pathway to it. This section presents some of those difficulties as puzzles. Their depiction as puzzles is deliberate, since the question of how far they are challenges to be overcome, rather than insuperable difficulties, depends in part on what further assumptions are made about contingent factors. While I will try to keep the discussion non-technical as far as possible, it needs to be prefaced by a more formal specification than hitherto of the conditions under which Coasian theory would predict that a consensus between states would be sufficient for Pareto improvement. Robert Inman and Daniel Rubinfeld suggest the following conditions: (a) low transaction costs in concluding or enforcing agreements; (b) widespread knowledge of the preferences of all states and thus of the ‘full range of possible trades’ between them; (c) states operate as good ‘agents’ of their ‘publics’ (d) gains can be divided, if necessary, through trades across issues (log-rolling) or compensations to losers on any one issue (side payments) (Reference Inman and Rubinfeld1997: 76–80). With this definition in mind, I suggest that a Coasian scheme of cooperation presents at least the following four puzzles.
Puzzle 1. The ‘non-arbitrary’ starting point
A much-discussed problem is that only a process of Pareto improvement – and not the point from which that process starts out – can ever be considered to be both chosen and fair from all points of view. To use Amartya Sen’s distinction, the Pareto criterion can only ever satisfy a ‘process’, and not an ‘opportunity’, standard of freedom: at best, it sets out standards for an undominated process or direction of choice, not for removing all possible forms of arbitrariness in starting points and distributions of ‘opportunities’ (2002: 505–506). Indeed, Sen remarks that, from many points of view, a ‘society can be Pareto-optimal and still be perfectly disgusting’ as for example where ‘the starvers cannot be made better off without cutting into the pleasures of the rich’ (Reference Sen1970: 22). In the case of the EU, the fairness of ‘starting points’ arises in relation to the rule that all new member states should accept the acquis communautaire in its entirety. We will come back to this.
Puzzle 2. Changing preferences and technologies
Other commentators point out that in a world of constantly changing preferences and technologies, a policy agreed by a consensus of states would be unlikely to remain Pareto optimal over time (Shapiro, Reference Shapiro1996). Random shocks, technological change, the partial endogeneity of preferences or, to put the point less technically, the very autonomy of what it is to be human and change one’s mind, will cause preferences in most polities to drift away from previously agreed Pareto points. As Douglas Rae puts it, this is almost certain to occur ‘unless the citizenry is dull as dirt’ or the public authority is ‘utterly clairvoyant’ (see Rae, Reference Rae1975: 1292). In a multi-state setting, and especially one that rests on indirect legitimation, there is the further challenge that governments, and even the institutional design of domestic polities, may change, with consequences for who can claim the authority to aggregate individual preferences into legitimately formed social preferences within member states. These factors mean that a policy endorsed by particular member states in the past may become unwelcome to them, raising, as we will see, the crucial question of how easy it is to restore them to their Pareto frontier.
On top of all that, the EU delegates significant powers to interpret legislation to the European Court of Justice with the consequence that it is not just preferences which drift over time, but the substance of the law itself. I will return briefly to this point.
Puzzle 3. Indivisibilities
The closer values are to being ‘continuous’ – such that they can be divided into ever smaller amounts – the easier it will be to ‘trade’ them in ways that allow actors to realize all possible Pareto improvements between themselves (see above p. 12). Yet many values are anything but continuous. They can only be enjoyed if ‘all or nothing’ choices are made to provide them in similar ways for large numbers of people. Kenneth Arrow puts the point thus: [Some] ‘actions being collective or interpersonal in nature, so must be the choice amongst them … The individuals in a country cannot have separate foreign policies or legal systems’ (Reference Arrow1973 [1967]: 123). He might have added that there are also limits to how far they can have separate market structures, separate institutions of macro-economic management, separate opportunities to breathe clean air, or separate welfare states. Not only are such choices affected by European integration, but, as we will see, the indivisibilities in question often seem to involve equally reasonable yet conflicting values.
Puzzle 4. Multiple equilibria
The search for Pareto improvement may produce multiple equilibria which leave all participants better off, while distributing the gain in different ways. The Pareto criterion will therefore be an insufficient condition for non-arbitrariness without some agreed means of choosing between the several solutions it may offer. Indeed, Rawls deftly links this to the first puzzle. If we have to choose between multiple ways of leaving everyone better off, we may want to take the initial distribution into account in allocating the increment (Reference Rawls1973 [1967]: 334–335). The (at least) two-dimensional structure of preferences on EU questions (Hix, Reference Hix1999) means that the EU is just the kind of arena where multiple equilibria are likely; and, in so far as Pareto improvement is itself thought to require some elements of majority decision to raise decision speeds and lower transaction costs, there is a risk that collective decisions will ‘cycle’ between an infinite number of possible equilibria in the EU’s two-dimensional policy space (McKelvey, Reference McKelvey1976). Discretionary and even manipulative choices of procedure (Riker, Reference Riker1986) will then be decisive in selecting outcomes, unless, of course, procedures are comprehensively specified in advance. The latter solution, however, departs from the hope expressed by Majone above (p. 6) that – in a Pareto-improving world – the main weight of legitimation in the European arena can be placed on the intrinsic qualities of policy outputs, thus lightening the need for procedural consensus. Moreover, indeterminacies associated with multiple equilibria may result not just from the dimensionality of single decisions but also from the sequencing of several (Pierson, Reference Pierson2000). Again, we will return to this.
Political justice
The puzzles set out in the last section allow us to take up Sen’s suggestion that greater use should be made of social choice in considering problems of moral and political philosophy. In the remainder of the article, I thus use the puzzles to evaluate the claim that the EU should operate as a Coasian scheme of cooperation against the standards I proposed above for the non-arbitrary exercise of political power, namely, political justice and democratic self-legislation.
I use political justice in the Rawslian sense of impartiality between competing, yet equally reasonable, conceptions of what is right and what is good (Rawls, Reference Rawls1993). It is, however, only possible to question whether a Coasian scheme of cooperation can ensure political justice if we first counter the claim that the only standard of justice needed under such a scheme is one where a consensus of states can secure what Brian Barry terms ‘justice as mutual advantage’ (1995; see p. 11 above). In what follows then, I first use the puzzles to question how far consensus bargaining between states can secure justice as a mutual gain. I then consider implications for political justice within member states.
Where better to start an assessment of how far an attempt to run the EU as a Coasian scheme of cooperation can ensure justice as a mutual advantage than with the problem of the starting point raised by the first puzzle? From the point of view of Coasian negotiation theory, it would only be in the unlikely event that all existing members would be better off constructing an enlarged EU on the basis of some alternative to the acquis communautaire that the EU would be likely to depart from its present practice of restricting new applicants to a ‘take it or leave it choice’ over existing policy obligations. Of course, new members will themselves only join if they are likely to be better off by doing so than remaining on the outside. But note that this says nothing about the distribution of the relative gain between new and existing member states. On top of that, the cost of non-membership – against which new members will have to decide whether they will be better off by joining – may itself be affected by the previous history of European integration. The very fact that outsiders have not previously participated as full members creates a risk that some EU decisions will have been made at their expense. Any tendency for members of an existing club to facilitate agreement between themselves by externalizing the costs of their own collective actions may even be aggravated where those governments happen to be democracies that are only accountable to their own national publics, and not to all those affected by their decisions (Grant and Keohane, Reference Grant and Keohane2005: 34).
Of course, defenders of a Coasian scheme of cooperation might well respond that it is precisely a focus on forward-looking and absolute gains that avoids burdening the legitimacy of the EU with any need to agree standards of relative gain or historical responsibility. Indeed, they would probably claim that a forward-looking focus on Pareto improvement can also solve the second puzzle: if preferences or circumstances change so that policies become sub-optimal for some member states, it should be possible to get the others to lift their vetoes on change by offering them ‘compensation’ or a ‘side-payment’ out of the gain to be had from returning the unhappy state to its Pareto frontier. One possible difficulty, though, with this response is that just as there may be some historical arbitrariness in where a state finds itself in the sequence of new entrants – and in the consequences to it of swallowing the acquis whole – so chance factors may influence whether they later need to supplicate change, or whether they, conversely, find themselves in the fortunate position of being able to extract a side-payment in exchange for agreeing changes to policies which have become unwelcome to others. If, indeed, we assume – in the Coasian approach – that institutions must sometime in the past have been ‘in an equilibrium’ where all actors have made best use of all information available to discover what is best for themselves, then unforeseen shocks are likely to be among the main reasons for departures from the Pareto frontier (Streeck and Thelen, Reference Streeck and Thelen2005).
A further difficulty is that it may not always be technically possible to compensate veto holders in ways that keep the acquis optimal (or even satisfactory) for all members. Scharpf has built an entire critique of the EU’s legitimacy on this very problem. From the empirical studies of several EU policies, he concludes ‘transaction costs are high… side payments and log-rolling are not always possible…and information on preferences is far from perfect’ (cf. the conditions for the application of CBT on p. 12). The result is that consensus decision rules can only be relied upon to align EU policies with the preferences of all member states ‘first time round’. Once policies are in place, ‘those same’ consensus ‘decision rules’ may create ‘joint decision-traps’ in which it is hard to change arrangements which may have become deeply undesirable to some member states (Reference Scharpf2007: 6).
However, Scharpf’s analysis does more than question whether a consensus of states can ensure ‘justice as mutual advantage’. In so far as member states cannot be quickly restored to their Pareto frontiers in the event of perturbations, shared policies will cease to be value neutral in the sense of being obligations to which they would all willingly consent in view of their own values and their own internal arrangements for legitimating value allocations. Indeed, in Scharpf’s view, in areas of policy that are central to the relationship between member states and their citizens, attempts to run the EU with rules that demand high levels of consensus result in precisely the opposite outcome to that predicted by Coasians: the external constraint of ‘change-resistant’ EU policies dominates domestic value allocation, rather than vice versa. Here, it is important to consider three of his claims in combination: (a) the EU entraps members in change-resistant policies; (b) the EU has pronounced ideological biases, notably towards negative rather than positive integration, to market liberalization, rather than social protection; and (c) the EU operates as a ‘compulsory negotiating order’ in which member states are legally and economically limited in how far they can pursue key objectives by other means if they feel constrained or disadvantaged by a EU policy.
Scharpf’s analysis is, however, open to a counter-argument as follows. It is an empirical question how large or small the problem he identifies turns out to be. Even if member states sometimes become entrapped in particular change-resistant policies, their overall membership may, as we put it earlier, remain Pareto improving in the round. If, on top of that, there are reasons to believe that compensation will work a good deal better within states than between them – because single states are solidaristic communities that can support redistribution to a degree the EU cannot – then each member state may be able to use its overall surplus from cooperation at the EU level to compensate those who lose out from any joint-decision traps or ideological biases in the European arena. Each participating society could decide to spend its cooperate ‘surplus’ in different ways. This might include among other things more spending on social and environmental policies. Only once account is taken of this secondary effect would it really be possible to conclude that even a EU that did negative integration and nothing but negative integration is ‘ideologically biased’. European integration could, in other words, be ideologically extremely narrow – and yet still expand the overall choice set of those with a wide range of different ideological preferences.
Whether this counter-argument really works requires us to turn to the third puzzle, namely, that of indivisibilities. It seems to me that there are sizeable indivisibilities in at least six choices affected by European integration: choices of (a) market structure, (b) environmental protection, (c) welfare states, (d) security provision, (e) collective definitions of political community, and (f) democratic institutions of rule (Offe, Reference Offe1998; Scharpf, Reference Scharpf1999; Bartolini, Reference Bartolini2005; Schmidt, Reference Schmidt2006). All are ‘gross choices’ (Dunn, Reference Dunn2001: 203) where it is hard to avoid at least some solutions that must apply more or less uniformly to those affected.
Now, each indivisibility is affected by integration in very different ways: (a) and (b) by the EU’s legal competence; (c) by the indirect effects of market integration and the free movement of persons and capital; (d) by EU-level coordination of policies that largely remain national competences; and (e) and (f) by Europeanizations of domestic arenas (Olsen, Reference Olsen2007: 227). Yet one thing is for certain: all six affect value allocations within member states via their participation in the obligations and decision rules of EU membership. Thus, the fairness of arrangements for dealing with the indivisibilities tests precisely that relationship which any theory of indirect legitimacy identifies as in need of justification, namely, whether the consent, control, and compensatory mechanisms available to member states are sufficient to justify those acts of coercion within their own arenas as required by the obligations of EU membership. Unusually, though, that challenge arises twice over in the case of indivisibilities. Decisions involving indivisibilities do not just need to be enforced. They are also coercive in the further sense that they involve a high level of pre-emption. It can be hard to make choices about them without drastically limiting what other choices can subsequently be made.
In answer, then, to the counter-argument to Scharpf, each additional indivisibility puts one more constraint on how far the overall package of outcomes in the domestic arena can be adjusted to offset any tendency for EU policies to privilege particular ideological positions or the status quo. Even that difficulty, though, is minor in comparison to another. All six indivisibilities are saturated with contrasting conceptions of what is right and good: of what rights are owed to others and of what are desirable ways of cooperating with them under shared mechanisms of social choice. It is thus unclear that it even makes sense to think of choices involved in (a)–(f) as tradable or compensable.
Now all polities, and especially liberal ones, face a challenge of how to make fair choices in the face of indivisibilities. That, it seems to me, is precisely the problem addressed by Rawls. What he identifies as the basic structure of society – rights and liberties, arrangements governing access to opportunities and life chances – is plainly an indivisibility. Indeed, at various points, Rawls identifies each of the further indivisibilities in (a)–(f) above as components of the basic structure (Reference Rawls1993: 257–288). Precisely because it is an indivisibility – that must in its essentials be the same for all members of society – Rawls points out that the basic structure of society can only amount to a fair scheme of cooperation if it can be endorsed from the point of view of opposing yet equally reasonable conceptions of the good.
This is not the place to discuss whether such an ‘overlapping consensus’ is fated to be an empty set (Richardson, Reference Richardson2003: 53) or, contrariwise, over-determined (Sen, Reference Sen2009). Rather, it is sufficient for our purposes to emphasize Rawls’s negative conclusion that one thing liberal polities cannot do is to privilege particular ideological positions or the status quo where (a) that has consequences for the basic structure of society and where (b) that amounts to an arbitrary selection between opposing but equally reasonable conceptions of the right and the good (Rawls, Reference Rawls1993: 193; see also Barry, Reference Barry1995: 82–85). Moreover, in Rawls’s view, partiality between a plurality of equally reasonable values involves a more fundamental failure of impartiality than distributive concepts of injustice (c.f. his earlier work, Reference Rawls1971). Whereas the latter are understandings of injustice from particular views of the good, the former is a failure to achieve impartiality between all possible views of the good (Reference Rawls1993, Reference Rawls2001). The problem will, moreover, be aggravated by the last of the public choice puzzles set out above. Procedural biases for or against opposing yet equally reasonable views of the good will be even less defensible if multiple equilibria over time mean that favoured positions benefit from increasing returns while disadvantaged ones face costs of switching between alternatives (Pierson, Reference Pierson2000).
Democratic self-legislation
In the last section, I doubted how far Coasian bargaining between member states can ensure political justice within them. I will now ask how far it can satisfy the other standard for the non-arbitrary exercise of political power set out above, namely, that of democratic self-legislation. Once again, I will aim to identify problems that may arise within member states on the assumption that a key test of any strategy of indirect legitimation is that it should not call into question the legitimating qualities of member states on which it depends.
To recall, the principle of democratic self-legislation requires that publics should be able to control the making, revision, and administration of the laws by which they are bound. Of course, it is a standard critique of attempts to legitimate the EU indirectly that the participation of national governments in the exercise of its legislative powers may fall short of what is needed for publics to see themselves as authoring the EU’s laws through representatives. James Bohman even argues that national governments practise a kind of ‘reverse agency’ (Reference Bohman2007: 70). Instead of controlling EU institutions on behalf of their publics, they use EU institutions to loosen controls their publics exercise over them. Others have fleshed out this critique with claims that national governments use European integration to reduce the scope of political competition (Mair, Reference Mair2005: 12), and to practise a form of executive domination, even executive appropriation, of legislative powers. National executive actors exercise lawmaking powers in the European arena (Weiler, Reference Weiler1997), even though they assemble together for purposes of intergovernmental bargaining and of pooling technical expertise. Those purposes are not easily combined with the procedural requirements of legitimate lawmaking – such as transparency and public justification (Eriksen, Reference Eriksen2009) – that may be needed if citizens are to see themselves as authoring their own laws through representatives.
Inman and Rubinfeld are surely correct to include effective agency among their conditions for the application of CBT to inter-state cooperation (see p. 12), if the latter is to be justified by its relationship to the governed. Although it is mistaken to see ‘doing what the people want’ as a necessary condition for democracy (since publics may themselves prefer representatives to use their own judgements (Plamenatz, Reference Plamenatz1973)), it is plainly a requirement of democratic self-legislation that any discretion afforded to lawmakers should be both controlled by publics and exercised on their behalf. Decision-makers cannot pursue goals that are peculiarly their own without being arbitrary in the manner Pettit defines the term above (Reference Pettit1997: 3).
However, I question how far attempts to secure indirect legitimacy through a Coasian scheme of cooperation can assure democratic standards of self-legislation, even where member state governments act as perfect agents of their own publics. Here, I make a number of assumptions that make no reference one way or another to whether governments operate as good or bad agents of their publics in the European arena. The first is:
(a) Indirect legitimacy presupposes that laws common to a group of states are justifiable according to those standards of democratic self-legislation which prevail in each participating state.
I now further assume that national standards of democratic self-legislation for normal lawmaking should (and in most member states more or less do) include the principles that:
(b) Present-day national majorities should be able to reverse the decisions of previous majorities, and
(c) In determining who should count as the majority in each Member State, decision rules should be more or less ‘symmetrical’ (Scharpf, Reference Scharpf2006) between those who favour status quo policies and those who would rather change them.
Although there may be much to the view that the ‘majority should only get its way with difficulty’, it is important that the difficulties are limited to obligations that the principle of democratic self-legislation itself puts on majorities, such as a need to justify why the preferences of others should be set aside in the framing of laws which are binding for all (Mill, Reference Mill1972 [1861]). Once those obligations have been discharged, it is hard to justify why the majority should not get its way (see also Bellamy, Reference Bellamy2007) on the terms set out in (b) and (c). After all, without principle (b), there is a problem of ‘rule by ancestors’; and without principle (c), there is a problem of political equality.
As seen, CBT purports to identify conditions where requirements such as (a)–(c) can be expected to align more or less automatically. However, this is not a robust finding. As shown in the puzzles above and by Scharpf’s joint decision-trap, it is only necessary to question the reality of one of the assumptions of CBT – absence of transaction costs – for it to be impossible for consensus-bargaining to satisfy all three requirements, (a)–(c), across time and across all member states. Where transaction costs constrain real-time adjustments of existing legal obligations so that they remain optimal, or even functional, for all member states, some members will find themselves obliged to enforce EU measures which would normally have been amendable under principles (b) and (c), while others may be unable to secure the domestic consent required by (a) for changes that would relieve problems faced by disadvantaged members.
Now, it is easy to think of reasons why majorities in particular member states should be held to obligations that have become unwelcome to them. Limits to ‘defection’ from agreements may be necessary to solve collective action problems. Moreover, European law, unlike domestic law, amounts to a shared body of commitments. It results from a decision-making process for which all member states have some responsibility. It produces policies upon which individuals from all member states are led to believe that they can rely on the formulation of their own life plans. Such arguments, however, are not easily made within assumptions of indirect legitimacy. They do not trace the legitimation of EU laws back to national standards of democratic self-legislation. Rather, they appeal to norms of consideration for others in other member states, and of shared historical responsibility for previous EU decisions. Their purpose is to justify the continued enforcement of EU obligations in spite of their possible incompatibility with standards of domestic self-legislation such as (b) and (c), not because of them.
Further difficulties follow from the substantial role reserved for judge-made law in the EU arena. Space does not permit their full consideration, except to make the fairly obvious point that only under assumptions of guardianship can judge-made law have a role in ameliorating, rather than aggravating, the difficulties which the foregoing puzzles – arbitrary starting points, preference drift, indivisibilities, and multiple equilibria – pose for the view that the historic consent of member states will always be sufficient to legitimate EU laws. The Court would surely need to exercise substantial judgement of its own to assess what would be fair in spite of these difficulties. To see such a guardian role as justifiable under a delegation from the demos of each member state would involve a huge substitution of judicial power for practices of democratic self-legislation.
My argument, however, is open to the fairly obvious objection that I have defined my assumptions too restrictively. In particular, it might be objected that even if it is hard to imagine indirect legitimacy without (a), there are good reasons for not always insisting on (b) and (c). After all, many democratic polities ‘constitutionalize’ particular decisions by putting outright legal and procedural restrictions on simple majority decisions which go well beyond the notion that the ‘majority should get its way with difficulty’. So why not justify the EU’s ‘change-resistant’ decision rules on just such a basis? But for such an argument to work implies two things that are by no means clear. First, that EU membership really can be understood as having been authorized in each member state according to the procedures that its own standards of democratic self-legislation require for the constitutionalization of some matters rather than others. Second, that the substance of European law – the kind of choices that it is used to enforce – is of a kind that would justify its constitutionalization. Although space does not permit a full discussion here, the obvious difficulty is one of stretching the justification of constitutionalization from the protection of democracy, individual rights, and cultural communities to the ringfencing of the economic and social choices which provide much of the content of EU law. The latter will often be specific selections from a wide range of different solutions that could have been chosen. Thus decisions which privilege the status quo may seem arbitrary (see especially Bellamy, Reference Bellamy2007)
Now that I have considered their implications for democratic self-legislation and political justice, the cumulative implications of the puzzles can be summarized by noting how they relate to different aspects of Coasian bargaining theory. The problem presented by the first puzzle could even arise under the Coase theorem in its original form as a theory of welfare maximization. Even if all parties end up by realizing all possible gains from cooperation, that cannot guarantee that the scheme of cooperation will be fair to the satisfaction of all, once the arbitrariness of starting points is taken into account. The remaining three puzzles – preference drift, indivisibilities, and multiple equilibria – become problems once we move on to Buchanan and Tullock’s reformulation of the Coase theorem as a theory of political bargaining. In part, this is because, unlike Coase, Buchanan, and Tullock assume transaction costs without which those three problems could be managed by compensatory payments. If, however, preference drift is not always compensable, it can no longer be guaranteed that collective choices entered into at time t by voluntary bargaining can still be justified at time t+1 on what Buchanan and Tullock hoped would be the libertarian and ethical grounds that those collective choices are also the individual choices of each actor. Applied to the EU, this implies limits to how far EU decisions previously agreed upon by democratic standards of self-legislation in individual member states will always remain justifiable by those same standards over time. That temporal difficulty will be most acute where the policies that are change resistant at the EU level involve either multiple equally reasonable solutions or, at the other end of the scale, indivisible choices, which can only be made, if at all, in much the same way for large groups of people.
Conclusion. Whither indirect legitimacy?
The section ‘CBT and indirect legitimacy’ showed how arguments for indirect legitimation work better where they can assume a Coasian scheme of cooperation. The section ‘social choice puzzles’, however, demonstrated that there are difficulties, even from a social choice perspective, in applying CBT to the EU. These include arbitrariness in starting points, preference drift, indivisibilities, and multiple equilibria. The sections ‘political justice’ and ‘democratic self-legislation’ then used the social choice difficulties to question how far Coasian bargaining between member states can ensure two conditions for the non-arbitrary exercise of political power within member states, namely, political justice and democratic self-legislation. Note, however, that it is the member state – the state as it is constrained by the political obligations and decision rules of EU membership – whose non-arbitrariness is called into question. This implies that if Coasian assumptions break down in the ways suggested here, they cannot ground a plausible theory of indirect legitimacy, which must be capable of showing how the state can retain enough of its legitimacy within a setting of European integration to also be able to do the work of legitimating the EU itself.
However, if the application of CBT to the EU fails, where does that leave indirect legitimacy? Can it be rescued or must it be abandoned? Scharpf, for one, maintains that the EU can only be legitimated indirectly, even though, for all the reasons rehearsed here, he also believes that attempts to run the EU as a Coasian bargain between member states will only lead to delegitimating joint-decision traps. I am, however, less certain that it is possible to doubt CBT and yet affirm indirect legitimacy. To see why, it helps to consider indirect legitimacy in relation to the EU’s enforcement structures. Although member states can be fined for non-compliance, held liable by private actors for the consequences of not meeting obligations established under EU law (Tallberg, Reference Tallberg2003), or just reminded of any shortcomings in their own implementation when they request new obligations of others, indirect legitimacy implies limits to how far EU institutions can sanction those very states it understands as being the EU’s own principals. In so far as it implies that member states retain monopolies on legitimate violence and on the ingredients of democracy and political community (Scharpf, Reference Scharpf1999), indirect legitimacy cannot justify coercive powers that have not themselves been delegated to EU institutions by its member states, and nor can it answer the question ‘what is the basis of political obligation to the EU?’ in any way that presupposes EU institutions can enjoy inherent, rather than delegated, democratic authority.
The difficulty of justifying why the EU should have ultimate sanctioning powers or why EU level majorities should prevail over national ones from within the assumptions of indirect legitimacy rules out two of the means by which modern polities deal with the problem that legitimacy depends in part on actors with confidence in the compliance of others. The importance of this problem cannot be underestimated. To the extent that political obligations are fashioned for the purposes of living and deciding together, it would be eccentric if they were not seen as losing their binding force once the prospects of others complying falls beneath the threshold needed to solve collective action problems (Føllesdal, Reference Føllesdal2004: 15–17). Even Rousseau – who of all people believed that legitimacy was a matter of moral obligation – hints at the need to anchor expectations of the compliance of all in order to secure the obligation of each when he remarks in the preface to the Social Contract on the importance of ‘uniting what right sanctions with what is prescribed by interest, in order that justice and utility may in no case be divided’ (1973[1762]: 165).
Cue CBT. In the view of its defenders, it can anchor expectations of the compliance of all in a view of what ‘right sanctions’ and of what ‘interest prescribes’, without exceeding the limits to indirect legitimacy set out in the preceding paragraphs. If indirect legitimacy presupposes limits to how far the EU can coerce member states, Coasians reply that that need not matter. Each member state can, at the end of the day, have confidence in self-enforcement by all the others. Where a pattern of cooperation is Pareto-improving in the round, no member states will have an interest in defecting from its obligations for long or wholesale. If indirect legitimacy presupposes limits to how far pan-European democratic majorities can trump national ones, Coasians would, once again, reply that this need not be a problem. A scheme of cooperation can draw its democratic sanctions from all its component democracies, and yet still operate optimally from the point of view of each. Under the right conditions, it can withstand the revolt of particular democracies by re-bargaining its own terms.
Since I suspect that CBT is the only theory of inter-state cooperation that can show how indirect legitimacy can be made to work within its own limits, I also suspect that if it fails, then so must indirect legitimacy. That would suggest little alternative to the EU establishing at least some of its legitimacy in a more direct relationship with the governed. Whether composite solutions that mix and match elements of direct and indirect legitimacy are possible, I leave to further speculation. What does, however, seem clear is that any solution would need to satisfy the same requirements for the non-arbitrary exercise of power – political justice and democratic self-legislation – discussed here. Three centuries after Locke, the challenge of political rule remains one of designing institutions that can deal with forms of arbitrariness in economy, society, and international relations without those institutions themselves being arbitrary. Indeed, the challenge of escaping the mischiefs of polecats without being devoured by lions is, arguably, even more critical to the design of institutions beyond the state than within it.
Acknowledgements
I would like to thank Fritz Scharpf especially for his comments, together with Christopher Bickerton, Daniel Gaus, Erik Eriksen, Andreas Føllesdal, John Erik Fossum and three anonymous referees.
