Vetoes, deadlock and deliberative umpiring: Toward a proportionality doctrine for power-sharing constitutions

Abstract

This article addresses power-sharing constitutions that include powers of veto wielded by discrete ethnonational groups. Such constitutional arrangements – seen, for example, in Northern Ireland and Bosnia – have often prompted severe deadlock, a problem that in turn threatens democratic functioning and raises the risk of renewed communal violence. We consider the use of ‘umpires’ of power-sharing constitutional systems to vet the use of vetoes and (potentially) to prevent their overuse or misuse. Power-sharing umpires are not uncommon in practice. However, as yet there is little scholarship evaluating how, in substance, power-sharing veto umpires should approach their task. Relying on deliberative democracy theory, the article outlines three forms of ‘deliberative agreement’ that, in principle, deeply divided groups may reach in the course of policymaking. It goes on to explain how existing proportionality doctrines drawn from federalism and rights cases can be imported into the power-sharing context to ‘scaffold’ these broad ideals. This approach, it is argued, may provide a more detailed, coherent and practically workable approach to umpiring power-sharing constitutions.

Introduction

Today it is widely accepted that democracy in deeply divided societies requires power sharing. ¹ While no two power-sharing constitutions are the same, the basic aim is ‘to prevent the monopoly, permanent or temporary, of executive, legislative, judicial, bureaucratic, military or cultural power’. ² To this end, power-sharing constitutions typically mandate institutions that are broadly inclusive or representative, and that are intended to give all of the main contending groups or communities in society a genuine say in matters of collective concern. As part of that, power-sharing constitutions often contain strong veto points. They do so to ensure that each community’s interests will be treated with equal consideration – that minority communities will not be persistently outvoted or ignored by majority communities concerned only to press their own advantage.

Vetoes are thought to be especially important when the interests at stake in a decision are regarded as ‘basic’, ‘fundamental’ or ‘vital’ to a community’s identity or to its members’ life chances. ³ Simply forcing a community to accept a decision that, for example, limits the use of its language in public life, or that imposes conditions on access to state-funded education or public employment that its members may find difficult to satisfy, could well endanger power sharing. Yet while vetoes can protect minority communities from adverse decisions of this sort, and hence, ideally, help consolidate their commitment to democracy, ⁴ they may also encourage communities – or, more usually in practice, the parties elected to represent them – to frame more and more issues, no matter how mundane, as vital interests. ⁵ Crucially, the more that parties respond to this incentive, the greater the chances of political deadlock or paralysis. ⁶ In some cases, governmental functions may be suspended and shared democratic rule may be replaced by some form of external or unilateral leadership.

The case of Northern Ireland offers a graphic, though far from unusual, illustration. ⁷ The power-sharing institutions established under the terms of the 1998 Belfast Agreement were suspended from 2002 to 2007 and again from 2017 to 2020. The most recent suspension began in 2021 and continued until early 2024, during which time Northern Ireland was essentially run by the Northern Ireland Civil Service with decisions also taken by the Secretary of State for Northern Ireland. ⁸ The factors at play in this case are both manifold and complex. Yet the Agreement’s veto provisions remain a major, and much-discussed, source of concern. ⁹

What cases of this sort show is that vetoes can disrupt democracy itself. ¹⁰ Vetoes can undermine democracy by stymying its capacity to get things done. ¹¹ Deadlocks under power-sharing systems may erode trust between groups, ¹² and even trigger the beginning or resumption of widescale armed violence. ¹³ Power sharing may be a necessary condition of democracy in deeply divided societies, but what is also needed is, as Lijphart termed it, a ‘spirit of accommodation’. According to this concept, leaders ‘must be willing and capable of bridging the gaps between the mutually isolated blocs’. ¹⁴ They must, that is, be willing to adopt a point of view that encompasses more than their own narrow interests, or at least be willing to compromise. But what would such willingness entail, and can specific institutions or their animating doctrines promote it, if not in the short-run then at least over the course of time?

Umpiring is a distinct constitutional design model in which an independent body or actor provides an authoritative answer to a dispute that the parties have not themselves been able to resolve. The umpire’s authority may derive directly from the constitution or from within existing constitutional arrangements such as an act of parliament. Umpiring may also have important international dimensions to it, not least in deeply divided societies. A case in point is the Office of the High Representative (OHR) in Bosnia and Herzegovina, an internationally sponsored institution responsible for overseeing the implementation of the civilian aspects of the 1995 Dayton Peace Agreement. The High Representative is not only the ‘final authority’ when it comes to matters of interpretation, ¹⁵ but also has the power to issue binding decisions when local parties are unable or unwilling to agree. ¹⁶ Umpires are therefore not unusual. ¹⁷ Yet they can be highly controversial. There are, for example, serious questions to be asked about the degree to which the OHR is accountable and to whom, and about the degree to which it has contributed (or otherwise) to state building, political efficacy or inter-communal reconciliation. ¹⁸ An umpire may help to clear a policy deadlock by adjudicating uses or abuses of a power-sharing veto. However, the creation of an umpire per se offers little reason, beyond mere hope, to expect that the parties will be satisfied with an umpire’s decisions or feel bound by them.

Whether an umpire can function as intended turns on at least two conceptually distinct, if related, sets of conditions: institutional-cultural and substantive. ¹⁹ Institutional-cultural conditions for umpiring include an official body capable of issuing independent decisions that enjoy broad societal acceptance and hence are widely perceived as legitimate. Substantive questions are, in an important sense, prior to this. What, in detail, should an umpire decide? What range of questions should it properly address? Perhaps more fundamentally still, what specific doctrines should an umpire apply when deciding if a veto is valid? Principles to guide these judgments remain elusive in both the literature and practice of power sharing – and may help to explain why umpires are often so contentious. ²⁰

Our focus in this paper is principally on the question of substance. We begin examining it by drawing from the literature on deliberative democracy. ²¹ This literature suggests how inter-group policy agreements might be achieved through deliberation, as opposed to more strategic and often deadlocked forms of decision-making. Vetoes should, we argue, be reviewed on the basis of their consistency with the interests of all those affected by their use. Achieving such consistency is difficult at best. Yet the deliberative literature suggests three broad forms of agreement – ‘public reason’, ‘integrative accommodation’ and ‘deliberative negotiation’ – that may be read as tailored to this goal. An umpire may, in the ideal at least, facilitate groups’ efforts to achieve policy agreement, or else invalidate vetoes and other conduct inconsistent with the pursuit of deliberative agreement in one or other of these forms. ²²

The difficulty with these idealised possibilities, however, is that in themselves they provide little to guide the putative umpire. We therefore ask how an umpire charged with vetting vetoes in a power-sharing system could translate the broad forms of deliberative agreement into practice. To do this we borrow from areas of contentious constitutional politics where umpiring is already standard. Proportionality doctrines are central to umpiring disputes over rights – which frequently have inter-group dimensions – and federalism – which invariably have these dimensions. We show, uniquely in the literature thus far, how the stages of proportionality testing instantiate the three distinct forms of deliberative agreement set out in the preceding part.

Left for another day is a full-length discussion of the institutional-cultural aspect of umpiring. We briefly stipulate here that an umpire should adopt any of a range of deliberative democratic designs of independent decision-making bodies (e.g., certain commissions or mini-publics) with lengthy records of deciding contentious issues fairly and with a minimum of political polarisation, including under conditions of deep communal division. ²³ However, our focus remains on the far more neglected questions of substantive doctrine. Without guidance, an umpire may struggle to know how to proceed – or proceed inconsistently in ways that raise legitimacy questions. Proportionality doctrines already common in related contexts may, we argue, be used to ‘scaffold’ deliberative reason in the context of power-sharing vetoes.

Of course, no prospective approach is guaranteed to succeed, even when grounded in analogous experience. Yet as Khaitan notes, ‘the practical payoff’ of ‘an idealised account … is that it helps us identify pathological’ patterns and establish ‘principles that constitutions should seek to optimise’. ²⁴ Moreover, in many communities, power sharing as currently practiced is all but guaranteed to generate further deadlock. Hence while there are no perfect constitutional designs, nor indeed perfect outcomes, there are at least better and worse substantive options for mitigating deadlock and its attendant risks.

Power sharing vetoes and deadlock

A standard assumption in liberal democratic theory and practice is that if voters disagree on which collective decision to adopt, and if each vote counts equally, the view of the majority ought to prevail. ²⁵ In divided societies, however, the fact that voting typically breaks down along communal lines means that some communities may find themselves persistently outvoted. ²⁶ Power sharing may give each community a seat at the table and the chance to have its say. But when a community’s vital interests are in play, and when the decision turns on little more than brute numbers, any ‘defeat will be regarded as unacceptable and will endanger elite cooperation’. ²⁷ Hence, vetoes are required to ensure that each community’s vital interests will be treated with due concern and respect. ²⁸ More broadly, vetoes are meant to ‘pacify’ parties, protect democracy and ultimately keep the peace. ²⁹

While power-sharing vetoes are meant to ensure that no community will hold the balance of power exclusively, the approach faces an array of challenges. A veto is an exceptional instrument and therefore not one to be used for a slight or trivial reason. Yet as already noted, vetoes can easily be abused, thereby increasing the risk of deadlock. ³⁰ On the one hand, vetoes are meant to be used only in the last resort, when other means of protecting vital interests have come up short. But on the other hand, parties have a real incentive to trigger them with ever greater frequency – irrespective of the interests of others or, indeed, the good of society as a whole. In that sense, one might say that vetoes are the enemy of compromise. They may inhibit parties from taking a broader view of decision-making and stymie the emergence of a stronger spirit of accommodation.

Even if a communal veto system works largely as intended, there are also enduring worries about the legitimacy and ultimate consequences of such a system. If certain named or ‘designated’ communities can veto decisions while other groups in society cannot, then the votes of the former group would seem to count for more. ³¹ It may be that ‘a hard confrontation with reality forces certain options on decision-makers in deeply divided territories’, including that of reserving vetoes for the main contending communities; however, the point of principle at work here still stands: some people’s votes carry greater weight than those of others. ³² Importantly as well, reserving vetoes in this way may actually make it harder to get beyond communal politics or to address important issues of society-wide concern. ³³

One ostensible alternative is to replace community vetoes with weighted-majority requirements. ³⁴ Thus the threshold for legislative decisions is set at a level above 50 per cent. Such a supermajority threshold may effectively require at least some support from each community, and potentially from communally unaligned groups. If well-designed, a weighted-majority requirement could potentially protect not just the vital interests of communities, but the interests of everyone in society, as well as manifesting a stronger commitment to political equality. Since (ex hypothesi) no party or coalition of parties from within the same community could reach the threshold on their own, it could also perhaps incentivise conciliatory behaviour, a spirit of accommodation and a focus on the wider public interest. ³⁵

However, weighted-majority voting at least sometimes reproduces the problems of communal vetoes, and to some extent amounts to a similar constitutional model. For instance, in Lebanon, the President (by convention a Maronite Christian) is elected by a two-thirds majority of the Lebanese Parliament. The President then appoints the Prime Minister (by convention a Sunni Muslim) with the support of the parliamentary Speaker (by convention a Shi’a Muslim). In turn, the Prime Minister appoints the executive or Council of Ministers which, according to the Constitution, ‘shall make its decisions by consensus’ or, failing that, ‘by vote of the majority of attending members’. ³⁶ Importantly, however, ‘[b]asic issues shall require the approval of two-thirds of the members’, including constitutional amendments, electoral law and the annual governmental budget. ³⁷

On the face of it, Lebanon’s power-sharing constitution contains strong conciliatory incentives. Yet while the two-thirds threshold has led leaders to look for support from communities other than their own, the parties with whom they seek to cooperate often turn out to be highly unpopular within their own community. The most recent impasse over the election of a president, and the controversy surrounding one candidate in particular, is a case in point. While Suleiman Frangieh had the backing of Hezbollah, his appointment was opposed by the largest Christian party, Lebanese Forces. ³⁸ Effectively, Frangieh’s appointment was vetoed by his own community.

Whatever the precise system of power-sharing, then, the risks of veto-induced deadlock are often very real. The Northern Ireland Assembly is a case in point. As noted in our introductory remarks, the Assembly has now been suspended on multiple occasions. While there are many factors at play, the legislature’s ‘petition of concern’ veto mechanism is a prominent consideration. ³⁹ The worry is not just the sheer number of petitions that have been tabled (159 to date), but also the range of issues to which they have been applied (the bill of rights process, the Irish language, flags, same-sex marriage, abortion, planning reform, traveller’s rights, etc.). ⁴⁰ While in the context of Northern Ireland, some of these issues can readily be construed as engaging vital community interests, the case for others is far less obvious or clearcut. Some petitions may even be construed as an abuse of the veto system. ⁴¹

The concern here has been widely acknowledged. ⁴² In response, as part of the January 2020 New Decade, New Approach reforms, the parties agreed that ‘the use of the Petition of Concern should be reduced, and returned to its intended purpose’. ⁴³ Moreover, they agreed to ‘publicly commit to tabling or supporting Petitions of Concern only in the most exceptional circumstances and as a last resort, having used every other available mechanism’. ⁴⁴ From January 2020 to January 2022, no petitions were tabled. ⁴⁵ Granted, in February 2022, the Democratic Unionist Party collapsed the executive once again in protest over post-Brexit trading arrangements, which in turn paralysed the legislature. ⁴⁶ Even so, the New Decade, New Approach document contained a further set of stipulations that are particularly germane to our topic. The parties agreed in 2020 not only to be more circumspect in their tabling of petitions but also that ‘a Petition must be accompanied by a statement of the grounds and rationale upon which it is being tabled…’. ⁴⁷ They further agreed that a ‘valid Petition of Concern shall trigger a 14-day period of consideration, including on any reports on whether a measure or proposal for legislation is in conformity with equality requirements, including the ECHR/Bill of Rights…’. ⁴⁸

Since no further petitions have been tabled, these stipulations have yet to be tested in practice. They nevertheless point to an approach to vetoes that calls on parties to clear certain additional, notionally deliberative steps when asserting vital interests. The reason-giving requirements stipulated might, for example, prompt the parties to set out grounds upon which a petition is being tabled, ideally by reference to broad rationales that others could agree to as well and that they might even rely upon themselves in the future. Moreover, triggering a 14-day period of consideration may make space for various kinds of negotiated compromise to occur (including, but not only, by reference to equality rights). Construed in this way, the New Decade, New Approach reforms may be read as containing the seeds of a politics of mutual accommodation based on inter-community deliberation and agreement. Yet one obvious problem with this reading of the agreement is that there is little guarantee that the parties will deliberate as expected. For instance, what is to stop communities from continuing to push claims that favour their own interests, or press their own advantage, without even attempting to cite broadly agreeable rationales? Moreover, what will happen when the parties’ views continue to conflict, even after the stipulated 14 additional days?

One possible answer to these questions is the introduction of an umpire. However, both literature and practice lack anything like a developed account of veto umpiring and, in particular, of the doctrines of adjudication on which such a body might rely. In the next parts, we first look at the normative theory of deliberative democracy in order to offer a general theoretical frame to understand how umpires of power-sharing vetoes could function. The breadth of these approaches is such that one might question the degree to which they can actually guide decision-makers. In response, we will later explain how proportionality doctrines, already common in related constitutional contexts, may be used to ‘scaffold’ deliberative decision-making in the context of power-sharing vetoes.

Three forms of umpired deliberation

By what means may umpires break up deadlocks that form between power-sharing groups on questions of veto? In this part, we lay out ideal answers, grounded principally in the normative theory of deliberative democracy. Deliberative democracy is an approach to democracy that stresses the importance of reason-giving in public-political life. As part of that, it stresses the importance of listening and reflection. Deliberation aims at mutual understanding. It also aims at agreement with respect to policies and procedures. Actual agreement may well be out of reach. But deliberative democracy nonetheless requires us to pay attention to one another’s arguments and respond to them rationally. Accordingly, the reasons that we give should be accessible and acceptable to others. This does not require us to set our own preferences or interests, positions or perspectives to one side. ⁴⁹ But it does require us to explain or defend them in general terms – for example, that we have special needs, that measures are needed to correct a historical wrong, or that our preferred policy would benefit society as a whole.

Admittedly, a great many writers on deeply divided societies, particularly those influenced by rational choice theory, tend to view constitutional agreements in terms of strategic bargaining. As Horowitz puts it, the ‘tacit assumption has been that, if the parties agree and can live with the agreement, they must have arrived at something like the market-clearing price’. ⁵⁰ But the growing interest in deliberative approaches to democracy in divided societies is at least in part driven by the vulnerability of strategic bargains to the underlying balance of power. ⁵¹ Just as importantly, strategic bargaining is largely unsuited to the challenges of independently adjudicating between competing claims – including claims concerning the use or misuse of power-sharing vetoes.

As noted previously, we do not dwell on institutional forms here, leaving a full treatment of institutional design to parallel work. Note briefly, however, that deliberative democracy and related literatures explore, often in considerable detail, options for institutional design of bodies capable of independent yet also representative decision-making. Representation can take many forms – such as formal, symbolic, descriptive and substantive. ⁵² Whatever form of representation is adopted, a deliberative democratic body aims to avoid seeing representation dissolve into polarisation and deadlock. Small bodies such as commissions and mini-publics can be designed to make debate and representation more deliberative, and have been shown as capable of settling policy disputes even in deeply divided and conflict societies. ⁵³

An umpire of power-sharing should be a standing body – ideally a permanent fixture of the constitutional order – empowered to adjudicate particular disputes over power-sharing vetoes. ⁵⁴ The umpire can be a court, but can also be a commission or mini-public. ⁵⁵ Commissions or mini-publics may be preferable, for three reasons. First, any institution can become politicised. Yet the politicisation of courts presents an outsized and perhaps unacceptable risk, given the wider role courts play (or ought to play) in upholding the separation of powers and rule of law. Secondly, while courts can be somewhat representative, ⁵⁶ the greater flexibility and capacities for representation of commissions and mini-publics count in favour of the latter, especially in divided societies containing multiple communal and non-communal groups and communities. Finally, commissions and mini-publics may more easily accommodate the inclusion of independent third parties capable of facilitating, enforcing and informing deliberation within the bodies – for example, actors drawn from foreign or international agencies, and academic experts. ⁵⁷

As already indicated, our aim is to outline what umpiring of vetoes should substantively entail – beginning with the broad kinds of interventions that an umpire may make, and finishing with three forms of deliberative agreement. As we have seen, power-sharing vetoes can be over-used or misused. An umpire may therefore be directed to overturn veto claims that fall outside of certain acceptable limits. We may call this ‘veto denial’. An alternative outcome is ‘veto affirmation’: the result of an umpire’s veto review may be that the umpire waves an appropriate veto through, leading in effect to a policy’s blockage.

Umpiring of power sharing may, however, diminish veto use overall. Umpires will presumably not let some vetoes through. Moreover, the umpire may offer what we can call ‘veto guidance’ over time, helping the parties to predict which other legislative vetoes will generally be valid, and which policies will fail by extension. An umpire should give reasons, either in the form of general principles or by making comparative assessments between policy options. These reasons may yield lasting guideposts as to which future vetoes will be affirmed or denied. The guidance may even coalesce into a set of precedents, akin to a court’s body of case law. The looming possibility that an umpire will affirm a veto may encourage a more conciliatory decision-making culture – one that prioritises accommodation over intransigence or deadlock, and that is therefore less likely to prompt excessive vetoes. ⁵⁸

Yet what kinds of reasons should umpires give? Note firstly that some reasons will concern whether an asserted veto rests on genuinely vital community interests. But what exactly is a vital community interest? The notion of an ‘interest’ can be straightforwardly described. What is in A’s interest is what makes for A’s good. It benefits or advantages A in some respect. Similarly, what is contrary to A’s interest disbenefits or disadvantages A. ⁵⁹ However, this description does not itself discriminate between interests of different type or weight. Earlier we suggested that vital community interests are interests that a community thinks of as basic to its identity (e.g., its right to speak its own language in public life) or as fundamental to its members’ life chances (e.g., equal education or employment opportunities). ⁶⁰ But while this suggestion may help to narrow the range of interests that can legitimately be described as ‘vital’, the fact remains that vital interests are likely to vary with the case or have a strong subjective dimension to them – which in turn opens up the possibility of over-use or misuse.

One possible response is to restrict the use of vetoes to a predetermined list of issues. (e.g., language usage, fiscal redistribution, constitutional amendment, etc.). ⁶¹ That list could be constitutionally enumerated or might be enacted through ordinary legislation and subject to periodic review. Yet while a restrictive approach can facilitate umpiring, it seems inhospitable to change. Some vital interests may be relatively permanent. But others may lose their force or importance over the course of time. ⁶²

It follows that an umpire’s determinations will depend on a close reading of the case to hand. Umpires must know the context. Beyond this, however, we must ask, what standards of validity should an umpire rely on when seeking to weed out unfair or unsupportable group claims? More particularly, how might an umpire manage the simultaneous and often incompatible claims that two or more groups in a power-sharing system invoke?

At least three key, conceptually distinct archetypes of deliberative democratic reasoning may (and in some cases in practice do) see parties in a collective decision-making enterprise find agreement: public reason, integrative accommodation and deliberative negotiation. These forms do not exhaust the possibilities for umpired deliberation, but they do cover considerable ground. As we stressed in the introduction, describing these ideals is not the same as claiming they are straightforwardly fulfilled. We occasionally witness deliberative forms of policy agreement in practice, even under otherwise deeply divided conditions. Yet such agreement is often incomplete, and seldom easily won.

Public reason

The public reason literature starts from the assumption that people are naturally free and equal. From this it follows that people should not simply seek to impose their political views on one another. On the contrary, since political decisions are mutually binding, the reasons on which they rest must be mutually acceptable. ⁶³ Or, in Larmore’s words, ‘[i]f the principles of political association are to be rooted in a commitment to equal respect, they must be justifiable to everyone whom they are to bind’. ⁶⁴ Therefore, especially when deciding constitutional questions or other basic matters of law or public policy, people should refrain from appealing to religious arguments or other ‘comprehensive’ views over which people are assumed to disagree. They should instead appeal to widely accepted political principles and values – such as freedom, equality, inclusion or respect – that anyone committed to democracy could reasonably be expected to endorse (whatever their more comprehensive commitments). ⁶⁵

The relevance of public reason approaches to deeply divided societies seems clear enough. A dominant majority community might claim, on religious grounds, that certain minorities should be excluded from state institutions or offices – that, for religious reasons, opportunities afforded to the majority should not be extended to the minority. But it is hard to see why minorities should be persuaded by arguments of this sort or regard decisions grounded in this way as mutually justified. To press ahead regardless would be to fail to treat minorities with equal concern and respect; in practice, doing so may well foment conflict.

The basic idea, then, is that decision-making should proceed from common ground – from shared principles or values, or other widely accepted considerations and standards (e.g., rules of inference or scientific reasoning). If the reasons that we give are not acceptable to others, we can try to find reasons that they will accept or, alternatively, moderate our views in the hope of arriving at a decision that is justifiable to all. ⁶⁶ On a public reason approach, an argument that is acceptable only from the perspective of one particular community, or that is cast in terms of principles, values or considerations that are not widely shared, has no place in democratic deliberation because it cannot form a basis of reasoned discussion. ⁶⁷

The need to justify one’s claims in terms of public values can have transformative effects. Rawls, for example, cites exclusionary comprehensive doctrines that many Catholics eventually declined to push for in the field of public policy. ⁶⁸ But while the adoption of public values may require cultural changes – Rawls hopes that groups and communities will ‘bend’ their cultural commitments so that they better support, or overlap in supporting, a democratic constitutional order ⁶⁹ – the obvious difficulty is that public values can themselves conflict. To take a commonplace example, freedom and equality can easily come apart when, in practice, freedom is defined as small government and equality is defined in terms of a welfare state.

Even when there is agreement on the relevant value, that value can be interpreted in radically different and conflicting ways. For instance, on the notion of sovereignty, Indigenous and settler groups may sharply disagree. Some who identify chiefly with the state will find offensive any suggestion that sovereignty does not belong exclusively to that state, or that it may be shared with Indigenous peoples. ⁷⁰ For their part, some Indigenous peoples, or their advocates, will find wrong or offensive any suggestion that sovereignty can only be defined in ways that, as they see it, solely serve the interests of those who seek to uphold the existing colonial order. ⁷¹

Public reason therefore can straightforwardly guide, on its own, only that small subset of cases that are effectively one-sided. For instance, a group that relies solely on religious rationales to seek to exclude another group from a public good, such as marriage recognition, pursues a position manifestly incompatible with public reason. ⁷² In such a case the excluded group has a relatively obvious and compelling claim against the exclusion. However, most inter-group policy disputes, not least in deeply divided societies, involve more complex competitions between group claims that may each, to some degree, be compatible with public reason. In these normatively multi-sided cases, the capacity for public reason to provide useful guidance, absent something more, is limited.

Integrative Accommodation

Ostensibly, integrative accommodation is a simpler form of agreement based on deliberation that uncovers common or compatible interests. ⁷³ Here the focus is not on the positions that people hold but on the reasons that underlie and explain those positions. To borrow an illustration, Fisher and Ury note how, at the beginning of the 1978 Camp David negotiations, the Egyptian and Israeli positions seemed diametrically opposed: the Egyptian delegation insisted on complete sovereignty over the Sinai Peninsula while the Israeli delegation insisted on retaining at least some control. However, focusing on interests rather than positions allowed a solution to emerge. What mattered most to the Israeli delegation was external security. What mattered most to the Egyptian delegation was historical ownership of the land. ⁷⁴ By reframing the conflict in this way, a solution was found. Egypt was given full sovereignty over the Sinai, but large portions of the peninsula were demilitarised, guaranteeing Israel’s security.

Integrative accommodation is a major theme in the negotiation literature. It is a form of negotiation that presupposes deliberation, though this is often left implicit. ⁷⁵ The focus on interests, as opposed to positions, is intended to uncover options for mutual gain – ‘win-win’ policies that can accommodate the interests of multiple competing groups rather than placing them in tension. ⁷⁶ As a first step, however, groups need to be able to articulate what it is they really want. In the integrative accommodation ideal, each helps the other to explore their interests and perhaps even to discover interests that they did not realise they had. As Mansbridge puts it, ‘one’s counterparts may come to the table with a position that one cannot accept, but genuinely interested questioning and close listening may reveal deeper interests that they have not fully articulated even to themselves’. ⁷⁷

Genuinely interested questioning and close listening are hallmarks of deliberation. They facilitate reflection and inventiveness, and the discovery of policy options for mutual gain. ⁷⁸ The policy that emerges may be one that, for instance, embraces overlapping Indigenous and state rights within a given space in order to accommodate each. Certain rights may be of minimal importance to one of the parties, but deeply important to another (e.g., preservation of Indigenous cultural sites; and continuation of traditional practices of hunting, fishing and resource use). ⁷⁹

Power-sharing constitutions can be understood, in part, in this light. Take again the 1998 Belfast Agreement. As O’Leary points out, Irish nationalists ‘endorsed it because it promises them political, legal, and economic equality now, plus institutions in which they have a strong stake, with the possibility of Irish unification later’. ⁸⁰ By contrast, British unionists endorsed it to ‘avoid the prospect of a British Government making further deals over their heads with the Irish State, and have some prospect of persuading northern nationalists that a newly reconstructed Union offers a secure home for them’. As such, the agreement marked a significant shift away from mutually exclusive positions (and all the violence that such positions brought) to a focus on underlying interests.

Deliberative Negotiation

The theory of public reason cannot, and was not intended to, apply to all types of cases. Moreover, win-win policy outcomes are not always possible or discoverable. Instead of securing mutual gains, some decisions turn out to be zero-sum. Some decisions entail choices that result in losses on all sides. For instance, Indigenous peoples’ traditional resource uses may be incompatible with state environmental standards (which may be more or less restrictive). ⁸¹ Zero-sum clashes may also arise where groups defined by ethnicity live in enclaves intended to protect their cultural coherence or safety. In such cases, it may be unrealistic for more than one group to use or claim ownership of the same space at once. A policy may seek to minimise conflict through land swaps in which some group members relocate to be closer to their ethnic kin; rather than win-win, however, this option may severely burden group members uprooted in the process.

Where irreducible differences between groups or communities arise, then, public reason and accommodation ideals can run short, and we must lean on less idealised alternatives. Even so, there may be more or less deliberative ways of conducting negotiations when some of the parties’ respective positions are irreducibly fixed or conflicting. In Warren and Mansbridge’s account of deliberative negotiation, the parties recognise that their interests are conflicting and that, to reach a decision, all sides must give ground. Yet rather than simply seeking to get the best deal possible for themselves, the parties seek to arrive at a decision that is fair to all sides. Each party is concerned to promote its own interests, but each also allows itself to be constrained by a shared concern for fairness. ⁸²

The concept of fairness admits of different definitions. However, a relatively deliberative process of negotiation, involving ‘relatively open conversations about underlying needs, interests, and constraints’, may nevertheless yield an agreement that ‘both (or all) sides in the negotiation perceive as fair’. ⁸³ In deliberative negotiation, policy outcomes are not simply consequences of power-based bargaining, ⁸⁴ but represent the genuine and diverse interests of the parties in a very particular way. Each side must still give up something of value – something that, in an ideal world, they would not have to give up. However, they do so for the sake of a policy outcome that all sides can regard as fair and that each regards as superior to the status quo ante.

There is empirical support for this ideal. Most deep inter-group division involves not just questions of identity but also questions of ‘distribution’ ⁸⁵ – for example, fiscal transfers from Flanders to Wallonia remain a central bone of contention in the Belgian case. ⁸⁶ The fact that questions of distribution are in play usually means that fairness will be fundamental. The leaders of the different communities may disagree on how fairness is best defined; at some level, they will probably seek to stack the deck in their own favour. But they will also know that any plausible (i.e., mutually acceptable) definition must include a concern for the good and interest of the other side.

Certain standard criteria for deliberative democratic procedures lend further definition to this ideal. Warren and Mansbridge focus on equality of representation, denoting the ability of groups or individuals to be included and heard rather than arbitrarily excluded. ⁸⁷ However, informational adequacy – which may refer both to a process that provides participants with relevant technical information, and to one that illuminates how policy choices affect members of various groups – is equally important. Among other things, it is often essential that people affected by a measure provide first-hand accounts of how it may affect them in ways others might not expect. ⁸⁸

These and other deliberative procedural hallmarks may prompt participants to adopt other-regarding outlooks (i.e., a concern not only for one’s own interests, but for others’ too), which aid fair outcomes. ⁸⁹ Fairness is never merely about being fair to oneself (though it is about that, too). It is also about seeing things as others see them, understanding their reasons, and weighing them in the balance equally with one’s own. In that sense, deciding what would be a fair outcome requires one side to give no more weight to its own interests, just because they are its interests, than it gives to the interests of the other side. The fact that the Belfast Agreement was reached at all must in small measure be put down to the willingness of the British and Irish governments to cooperate and, in some moods at least, to play the role of ‘neutral broker’; ⁹⁰ that is, to ensure that the agreement was fair to both communities and not disproportionately burdensome on either.

In sum, a power-sharing umpire should, in line with public reason, review and reverse any veto that pursues group claims that cannot be justified on terms that others could reasonably be expected to accept. The umpire should also seek out win-win policies that are likely to accommodate all groups. Alternatively, on matters where group claims unavoidably lie in tension, the power-sharing umpire should facilitate deliberative negotiation, and deny any veto that does not fairly consider the range and gravity of each group’s competing claims. In substance, these options may help to remove at least some power-sharing policy deadlocks. Again, these are ideals, or likelihoods at best; no deliberative process can be guaranteed to fulfil them.

In the next part, however, we turn to how deliberative agreement may be concretised. As described in formative works such as those of Rawls, Mansbridge and Warren, and others, deliberative agreement remains too indeterminate to steer practical decision-making. How exactly should an umpire try to invoke public reason, accommodation or deliberative negotiation in a case before it? The three distinct forms outlined above can orient our deliberations, but they remain underspecified, especially, in the present context, for the still largely hypothetical role of power-sharing umpires. To understand how umpires have carried out their tasks in analogous cases, we turn to the contexts of federalism and rights, where, we contend, proportionality testing often instantiates the different forms of deliberative agreement, adapting these from abstractions into more specific and workable umpiring doctrines.

Using proportionality tests to ‘Scaffold’ umpiring

In many cases on umpiring rights or federalism disputes, the forms of deliberative agreement above are implicitly at work. We have already specified that a trusted deliberative democratic body – ideally a commission or mini-public, but perhaps a court – is also a condition-precedent for effective umpiring. Again, however, our focus is on the substantive doctrines that such a body may employ to scaffold deliberative agreement. Such doctrines should aim to translate the broad normative directives to deliberate into more tangible, stable and detailed tests for umpiring particular disputes in a power-sharing veto system.

According to Alexy’s influential theory, the reason we need proportionality reasoning is that, unlike binary rules, rights are principles with ‘optimization requirements’: they are ‘norms which require that something be emphasised to the greatest extent possible given the legal and factual possibilities’. ⁹¹ According to Alexy, ‘principles, each taken alone, always comprise a mere prima facie requirement’. ⁹² In turn the ‘need for proportionality testing arises from the nature of constitutional rights as principles’, as such principles call for a determination of the ‘appropriate degree of satisfaction of one principle relative the requirements of another principle’. ⁹³

This is important background for the discussion of this part. We aim to show that various steps of proportionality resemble assorted forms of deliberative agreement, and to propose how the proportionality practices of federalism and rights can be transplanted, with modification, into the umpiring of power-sharing vetoes. The forms of deliberative agreement generally eschew binaries and instead adopt distinct kinds of weighing of principles, often by degrees. That the steps of the globally standard legal test for proportionality instantiate each of the three deliberative agreement types outlined previously is no coincidence, but a case of convergent evolution. Over more than a century, proportionality doctrines developed their discrete steps as logically distinct forms by which to accommodate or balance competing principles, values or group interests. ⁹⁴ Each of the forms of deliberative agreement we will highlight involve ways in which the claims of groups can be managed.

As Jackson observes, both federalism and rights cases often invoke versions of the internationally standard legal test for proportionality. ⁹⁵ Both are, as well, at least partly analogous to disputes over power-sharing vetoes. The federal division of power into coequal sovereign spheres is conceptually similar to the empowerment of separate yet equal communal groups or segments in executive or legislative power-sharing. (Indeed, a federal or similar arrangement can itself form part of a power-sharing constitutional design, as in, e.g., Belgium, Bosnia and Iraq. ⁹⁶) Just as the need to manage group differences is prominent in veto disputes, then, this need also arises in federalism cases addressing conflicts between federal and state power. ⁹⁷ An element of managing group differences is also inherent in many group-held rights or freedoms (e.g., equality, language, self-determination, self-governance and association), and indeed arises wherever discrete litigant groups’ claims compete (e.g., in hate speech cases affecting minority groups’ statuses, dignity or security). ⁹⁸

Proportionality testing can reshape – that is, block, pare down or revise – governmental measures that cause groups’ rights (or interests) to clash with one another. Proportionality testing may sometimes expose the manifest weakness of a rationale behind a measure, leading to the measure’s invalidation and handing a clear ‘win’ to the dispute’s plaintiffs. More commonly, however, proportionality testing manages the more complex, genuinely multi-sided cases in which multiple litigants pursue claims that are, at least to some degree, each well-justified and consistent with public reason. In this set of cases, proportionality may yield nuanced outcomes, such as complex accommodations or compromises.

The proportionality tests of both federalism and rights doctrines often feature a common sequence of steps, beginning with a threshold question: (a) Does a measure have a legitimate objective? A government generally has the onus of demonstrating that the objective of a measure is legitimate. ⁹⁹ In rights cases, including cases on equality, ¹⁰⁰ a measure can be illegitimate on its face if, for example, its main aim is to exclude a given group from public life arbitrarily, or to enable their arbitrary domination or abuse. In cases on federalism, an illegitimate measure may be one whose sole objective is to empower or enrich one jurisdiction at the expense of others (e.g., protectionist state tariffs violating federal free-trade provisions). ¹⁰¹

If a measure’s objectives clear the threshold test of legitimacy, then the inquiry turns to steps assessing the relationship between the measure’s ends and means. Three further questions arise at this main stage. (b) Is there a rational connection between the measure and its objective? In other words, is the measure drafted in a way that will likely lead, in practice, to the fulfilment of its objectives? (c) Is the measure least-restrictive? A least-restrictive measure is one that achieves an objective, to a specific degree, by means that are less burdensome than any reasonable alternative. This step may compare the existing measure with those in place in other jurisdictions, or with hypothetical less-restrictive alternatives. ¹⁰² (d) Are the weights of the objectives and of any burdens imposed by the measure in balance with one another? In this step, the objectives and burdens of a policy directly compete. ¹⁰³

As noted, we adopt the view that proportionality testing provides a normative framework to impose distinct forms of deliberative agreement on governmental measures. We are not the first to do so. According to Kumm, the proportionality test ‘provides a structure for the justification of an act’ through deliberation based on public reason. ¹⁰⁴ But while Kumm’s focus is on deliberation as public reason, our focus includes public reason but also ranges more broadly. Proportionality testing, as we will shortly see, concretises each form of deliberative agreement outlined above.

It may seem surprising that we identify proportionality as potentially instantiating forms of deliberative agreement, given the element of mutual consent or volition that agreement normally connotes. Many scholars describe proportionality as central to the imposition of a culture of scrutiny and justification of official action. ¹⁰⁵ Agreement may not therefore seem to be part of the standard picture of proportionality, in which acts of governments are generally understood as suspect, and the officials responsible for the acts bear the onus of marshalling evidence and arguments for their justification.

Yet even though the umpire has the last word on the substance of the case, the parties’ continuing volition may be potentially found in forms of participation leading up to an umpire’s decision. As Kumm observes, umpires such as courts can ‘compel public authorities into a process of reasoned engagement’. ¹⁰⁶ What this can mean is that an umpire facilitates engagement between the parties in ways that promote deliberation and reduce certain recurring barriers to agreement. ¹⁰⁷ The procedures and decisions of the umpire can, for instance, highlight public values that groups share in common, ¹⁰⁸ identify how reforms premised on accommodation may be possible, or lead the parties toward outcomes based on fair compromise.

We show next how proportionality testing instantiates these discrete forms of deliberative agreement. We also show how such testing can be adapted for use in the umpiring of power-sharing vetoes. This adaptation may occur via a formally enacted policy involving legislative amendment or even changes to the provisions of a bill of rights, power-sharing agreement or power-sharing constitution. Alternatively, as some courts have, ¹⁰⁹ an umpire may choose to self-impose these structured proportionality practices.

Legitimate objective

The threshold proportionality step, which as we saw tests the legitimacy of policy objectives per se, relatively straightforwardly instantiates public reason. Under Canada’s Charter, a governmental objective must be ‘demonstrably justified in a free and democratic society’. ¹¹⁰ The European Convention on Human Rights allows limits on certain rights if they are ‘necessary in a democratic society in the interests of public safety, for the protection of public order, health or morals, or for the protection of the rights and freedoms of others’. ¹¹¹ In Australia, an objective must be ‘compatible with the maintenance of the constitutionally prescribed system of representative and responsible government’. ¹¹² Each of these broad formulations has been read as committing the jurisdictions in question to bedrock principles including individual dignity and peaceful coexistence among groups, ¹¹³ and each resonates with public reason’s focus on inclusion and mutual respect, and managing or tamping down communal conflict. ¹¹⁴

As for what policies may specifically fall short at the legitimate objective step, illegitimate objectives especially may stem from group animus or political partisanship. Thus, some comprehensive doctrines offer rationales for policies that elevate one group at others’ expense (e.g., same-sex marriage exclusion), ¹¹⁵ based for instance on tradition, divine inspiration or naturalistic fallacy. ¹¹⁶ Moreover, manipulated electoral laws can entrench an incumbent party or societal group in power. ¹¹⁷ In either type of case, an umpire may invoke the threshold step to exclude the objectives outright and invalidate a policy or law.

There is little to prevent the adoption of legitimate objective testing to power-sharing veto umpiring. Indeed, in deeply divided societies, executive or legislative factions may be especially prone to using harsh measures, such as brazen electoral manipulations, to express dominance or group animus. Legitimate objective testing therefore may be a useful tool by which an umpire of power-sharing can block extreme policies that lack any plausibly legitimate objectives.

To be sure, cases where a policy is invalidated at the threshold stage due to incompatibility with public reason will likely be rare. In Northern Ireland, for example, only a few veto matters involved just one side making claims that clearly clashed with public reason. For example, the Democratic Unionist Party opposed same-sex marriage based on religious and traditionalist reasons that others could not reasonably endorse. Other matters were not similarly normatively one-sided; most instead provided rationales at least plausibly compatible with public-reason. Such complex multi-sided conflicts between groups are common in rights and federalism cases. For example, a law that gerrymanders an electoral map may sometimes be justified – or may at least cloak its partisan motives – based on notionally legitimate rationales such as enhancing underserved rural regions. ¹¹⁸ We will see shortly how later steps of proportionality testing address subtler or more insidious cases such as these.

However, note a doctrinal complication that arises when we turn from proportionality in rights and federalism cases to proportionality in power-sharing veto cases. In rights and federalism cases, the onus is on a government (or branch thereof) to abide by basic guarantees when adopting a new policy. By contrast, vetoes are the actions of parties within government that act to oppose other parties. Thus, the usual unilateral onus on ‘government’ to show legitimate objectives cannot apply. The onus must instead be shared between the party (or parties) proposing a measure (i.e., legislative bill or executive policy) and the party proposing to veto the measure.

Keeping the unilateral onus intact could lead to the perverse consequence of deepening, rather than calming, divisions. We cannot be certain, in advance and as a general rule, whether either the vetoed policy or the veto itself will lack any legitimate objective. For example, same-sex marriage exclusion may arise either where a party introduces a bill to bar non-heterosexual marriage, or where another party seeks to veto a bill meant to open marriage up to non-heterosexual couples. A ‘dual onus’ that reviews both the veto and the measure being vetoed therefore would be ideal as it could provide standards of public reason that apply evenly to all parties involved.

The umpire’s dual onus test for legitimate objectives would, however, mark a departure from most rights and federalism cases. In rights cases, any right invoked is, of course, assumed to be legitimate; only the governmental policy in question is suspect and must be justified. Similarly, in federalism cases, whichever government (state or federal) introduces a policy typically bears the responsibility of justifying the policy’s legitimacy. Dual onuses are therefore absent at the threshold step in rights and federalism cases. Notably, since the dual onus approach in veto cases would subject multiple groups’ normative claims to scrutiny, the approach would build on more basic approaches of public reason, which (as we saw) have little to say about how two or more groups’ legitimate interests or values might be analysed together.

We turn next to the three remaining steps of proportionality testing – each of which also involve two-sided analyses, and in the latter two steps go further to include relational analyses of group claims vis-à-vis one another.

Rational connection

This second step effectively focuses, again, on public reason. It identifies a range of relatively subtle cases in which laws have apparently legitimate objectives but do not give effect to these objectives in practice. Such laws cannot be justified to all, or in ways that all sides could in principle accept, and indeed may be animated by ulterior, illegitimate objectives.

For example, a policy barring entry of travellers into a state during a pandemic may be rationally connected to the objective of stopping virus spread; the policy’s breaches of constitutional free trade or free movement provisions may therefore clear this step of the proportionality test. Addressing border closures imposed by the state of Western Australia, the High Court of Australia found that there ‘can be no doubt that a law restricting the movement of persons into a State is suitable for the purpose of preventing persons infected with COVID-19 from bringing the disease into the community’. ¹¹⁹ However, hypothetically, the same policy might have become irrational later on, once the virus had spread widely within the state. In this case, the state would appear to be using a legitimate (notional) policy objective to give cover to the illegitimate (actual) objectives of the law. Pandemic border closures were indeed often criticised as populist, implying that their true inspiration was to sow division and exact political gains by setting insiders apart from outsiders. ¹²⁰

The essence of rational connection testing is thus the vetting, often in trial courts, ¹²¹ of concrete evidence in order to scrutinise claims that a policy pursues a legitimate end. Evidence from the physical and social sciences, such as expert testimony and published scholarship presented as documentary evidence, is often part of this step. ¹²² In the light of evidence-based review, a government’s justification of a policy may not stand up to scrutiny.

Turning to prospective cases of veto umpiring, in principle this step can operate in the same way to diminish inter-group tensions. (Note that, consistent with the dual onus in the threshold step above, legitimate objectives asserted by both the policy-proposing and the vetoing parties would be assessed here.) For example, a measure that supports a minority language by barring other languages on public signs may have a legitimate objective, but may clash with expressive freedoms asserted by speakers of other, deemphasised languages. Language policies, as already indicated, have been frequent veto targets in Northern Ireland (albeit in some cases indirectly). They often arise as well in other multilingual jurisdictions (e.g., Canada and Belgium). ¹²³ Rational connection testing can examine research evidence on the ability, or otherwise, of language measures to promote a minority language in practice. The evidence may also detail the centrality of language to the identities of speakers of the excluded languages. Conversely, however, the inquiry might reveal that the measure does not effectively pursue its own objective. Given this, the measure should be set aside as an unnecessary barrier to inter-group agreement.

Of course, the evidentiary analysis may instead find support for claims about a measure’s efficacy, leading to the decision that the measure should be upheld at this step. For instance, in Ford v Quebec, the Supreme Court of Canada found French-language signage laws to be rationally connected to their objectives. ¹²⁴ The Court relied on ‘general studies on sociolinguistics and language planning and articles, reports and statistics indicating the position of the French language in Quebec and Canada’ – all of which, according to Quebec’s government, ‘justif[ied] the language planning policy’. ¹²⁵

Least-restrictive means

Even if the measure has legitimate objectives and rationally serves those objectives, the next step assesses whether the measure is appropriately tailored in light of its deleterious impacts on others. The umpire may find the measure invalid if it is unduly restrictive of a given group’s rights or interests. For example, in Ford, despite finding the French-language measures to be both ‘serious and legitimate’ in their objectives and ‘rationally connected’ to those objectives, the Court ultimately found the infringement of freedom of expression to be more than was necessary to achieve the objectives. ¹²⁶

From the perspective of deliberative agreement, we may view least-restrictive means testing as important in two respects. The first again relates to public reason. An umpire may consider that a policy exceeds its own legitimate objectives. As Kumm puts it, a law might be ‘opportunistically’ designed to be disproportionate. ¹²⁷ Jackson similarly observes that in federalism cases a disproportionate law ‘may be a law not really designed for one asserted purpose but to sweep more broadly or in other directions’. ¹²⁸ Those ‘other directions’ may be ulterior ends incompatible with the exercise of public reason.

Least-restrictive means testing therefore scrutinises whether, despite a policy having legitimate objectives, an illegitimate objective also is at least partly at work. ¹²⁹ For example, in federal free trade cases, environmental objectives may be cited to justify curbs or tariffs on polluting industries, or social objectives may be identified to regulate moral or psychological harms (e.g., sports betting). But the umpire engaging in factual assessment of the policy may determine whether, and especially to what degree, the policy is actually justified. ¹³⁰ The policy’s excess – any component beyond what is strictly needed to achieve a legitimate objective – may be a result of poor drafting. But it could also indicate an ulterior, illegitimate aim incompatible with public reason (e.g., an effort of one part of government simply to arrogate power to itself). Least-restrictive means testing thus preserves interests and objectives that are legitimate in a public reason sense, but uses evidentiary testing to flag overreaching policies that may be based in part on illegitimate (e.g., comprehensive and exclusionary) doctrines.

This aspect of the least-restrictive means step is straightforwardly adaptable to cases on power-sharing. For instance, a power-sharing group may include some members who rely on violence to pursue political aims. In response, a party may propose harsh security policies, such as limits on group members’ movement into each other’s enclaves. The policy is per se justifiable on grounds of public safety – a public value shared by all. But while the ends may be sound, the means may include excessively harsh restrictions prompted by hasty overreaction or group animus. The umpire’s least-restrictive means test may, then, fine-tune the impact on rights by tying a measure to a dynamic risk assessment – for example, by limiting the measure’s application to times of heightened tension (e.g., nationalistic holidays and commemorations). In this way the proportionality test can, to an extent, preserve other important public values.

However, least-restrictive means testing is also potentially concerned with other forms of deliberative agreement, beyond public reason. In particular, such testing also prompts a search for policies that integratively accommodate the interests of multiple groups. For example, in some cases on federalism, this step can reveal policy options that do not diminish interests of either the federation or of any subnational state. Returning to the example of pandemic border closures, rather than block entry into a state altogether, the state could instead have regulated entry by subjecting travellers to a period of quarantine. This alternative policy may have preserved both the objective of preventing virus spread and the freedom of movement across state borders. If (and only if) this alternative is equally effective as outright border closures, then the test may highlight a win-win policy. ¹³¹ Similarly, in rights cases, as Levy and Orr point out, while a policy such as a height requirement for firefighters – which serves as a proxy for physical suitability to the role – might violate the equality of female firefighters who tend on balance to be shorter than males, a more accommodative policy would measure more direct markers of fitness such as strength and endurance. The new policy would reduce discriminatory impacts while still achieving its intended ends. ¹³²

Integrative accommodation could similarly apply to a range of veto cases. For example, where a single historic site is a centre of worship for two groups in tension with each other, options for regulating the space include granting each group separate exclusive areas or times to access the site. Such win-win solutions are imperfect in practice; they may even at times become focal points for violent confrontation. ¹³³ But they are generally preferable to less nuanced alternatives, such as those where groups view their claims as exclusively legitimate and push for their own community’s sole access to the entire site at all times.

To be sure, as Levy and Orr have observed in relation to rights cases, the kinds of accommodation we see here do not always occur. For instance, decision-makers who apply proportionality tests too rigidly, without a sense of the value of accommodation, may artificially separate out a measure’s objectives from its means. ¹³⁴ Proportionality tests typically take for granted that a right has been burdened and mainly focus on whether the burden is justified. This can preclude integrative reasoning, as is clear from cases where courts struck down measures aiming to enhance political deliberation (e.g., publicly-funded political advertising and campaign finance limits). The values of expression and deliberation in such cases may in theory inform and enhance one another, rather than necessarily conflicting. ¹³⁵ Yet if proportionality tests are too quick to find that a breach of a free speech guarantee has occurred, they may leave little room for such integrative arguments.

Thus, a risk amid umpiring is that an umpire will apply a relatively rote and conceptually narrow approach that overlooks the possibilities of accommodation. Accommodation may not simply happen on its own but may require conscious application: an umpire’s embrace of substantive decisions instantiating deliberative agreement. Concretely, in least-restrictive means testing the umpiring body either should not unduly dismiss any accommodative measures before it or should unfavourably compare non-accommodative measures before it with more accommodative alternatives. Hence the validity of the measure would be a function, in part, of the umpire’s reading of how accommodative the measure is, especially in comparison with real or hypothetical alternatives.

Balancing

As noted previously, however, many policy conflicts in any case cannot be readily resolved in ways that accommodate all groups’ interests. ¹³⁶ China’s sovereignty claims over vast swathes of the South China Sea appear to be incompatible with those of Brunei, Indonesia, Malaysia, the Philippines, Taiwan and Vietnam. In such cases, the main part of the dispute is typically zero-sum, or at least is often framed as such. Many policy conflicts in power-sharing systems have a similar zero-sum structure, including disputes over the sharing or transfer of revenue (as in the case of Belgium mentioned earlier) or the drawing of intrastate boundaries (as in the case of Kirkuk in northern Iraq). ¹³⁷ An umpire needs a way to address this freighted class of cases.

The balancing step of proportionality may, in principle, enforce a form of deliberative negotiation in which all parties, weak or strong, may see their interests robustly heard and considered on merit. Umpiring is potentially a useful replacement for, or addition to, negotiations conducted by the parties themselves. On their own, the parties are not likely to apply ground rules and procedures of equal inclusion and robust provision of information, nor to achieve anything like other-regarding perspectives, especially when there are large power imbalances between them. Using umpires may alter disputes in spaces (e.g., executive, legislative and public forums) that are otherwise largely deliberatively unregulated, and where deadlocks corrosive of trust are common.

However, to avoid simply replacing the parties’ decision-making role altogether, the umpire’s procedures and governing doctrines must robustly include ways of taking representations from the groups affected – as is common within the modern practice of rights or federalism proportionality testing within courts and commissions. The main groups affected tend to have extensive opportunities to present their positions, alongside assorted third-party intervenors. ¹³⁸ These representational and participatory methods in principle allow the parties to influence decisions affecting them.

Precise information on the matters subject to balancing is typically drawn from governmental or published research and from the evidence gleaned from the parties themselves. ¹³⁹ These sources can expand the range and nuance of interests to which the umpire is exposed. Evidence may be particularly persuasive if based on scientific, economic or anthropological research, ideally with a longitudinal dimension. For instance, cases on Indigenous rights often rely on copious research data about disenfranchisement and social exclusion over time, or about the richness of historical Indigenous connections to specific areas or cultural practices. ¹⁴⁰ Yet, as every case and every party is unique, it is also critical that an umpire hear different perspectives directly, including how individual members personally understand the dispute and its stakes.

Kulenovic states that one of the steps that the Bosnian Constitutional Court takes in its umpiring role is determining whether a vital interest would be injured, and argues that this step is entirely possible to place in standard legal reasoning, as it concerns a question of the prohibition of collective discrimination. ¹⁴¹ While we agree, we stipulate that the umpire’s decisions must ultimately be driven by the values, rights and interests of the parties themselves. It is beyond the capacity of a notionally objective umpire to sort out, unilaterally, many of the questions raised in a veto context. For instance, the question of vitalness must ultimately be answered by the members of a group themselves, who may or may not count a matter as foundational or linked to their collective identity. The two noted sources of evidence may help the umpire to determine which interests are genuinely vital to the group. The umpire may then weigh these and other interests against each other, ideally as a relatively passive facilitator of deliberative negotiations driven ultimately by the parties themselves. ¹⁴²

To be sure, this model is open to the long-running criticism that umpires may, in reality, take important choices about values and interests into their own hands, and therefore out of the hands of democratic (including power-sharing) actors. ¹⁴³ Umpires may substitute their own views, then, for those of their notional constituents. However, as Möller observes of judges who ‘fail to live up to’ a certain standard of proportionality testing, ‘this is the judges’ fault, not proportionality’s’. ¹⁴⁴ Indeed, as we said of integrative accommodation above, for an umpire to fulfil a normative ideal may require, in some measure, the umpire’s conscious efforts to do so. The umpire should allow the parties to air their own viewpoints, which should in turn dictate how the umpire ascribes weights (e.g., degrees of vitalness) to the values, rights or interests in tension. Under the umpire’s facilitation, the power-sharing groups may in effect be forced to reckon with the full range and gravity of each other’s claims. ¹⁴⁵ Ultimately, though, it must be the umpire, rather than the parties, that determines the outcome. ¹⁴⁶ This is a necessary expedient to help move the parties’ interactions away from grandstanding and ‘heat[ed]’ interactions ¹⁴⁷ and deadlock, and into the more controlled conditions of deliberative negotiation.

Conclusion

In examining power-sharing vetoes, we have, admittedly, focused on only one element of a power-sharing constitution. The different parts of a power-sharing constitution are likely to be mutually implicating, which calls for a more holistic account than we have offered here. ¹⁴⁸ Nevertheless, we have proceeded on the basis of a belief that institutions matter. In particular, the institutions within which people deliberate about contentious matters of law or public policy may impact the outcomes that emerge. A key concern in deeply divided societies in general, and power-sharing democracies in particular, is the risk of veto-induced political deadlock. Sometimes it may make sense to take decisions out of the main protagonists’ hands and give them to independent umpires.

This is not a new thought. Nor is the thought that decisions need to be justified to all, including decisions to veto proposed legislation. But on what basis should they be justified? What principles or values should be involved? In this article, we have sought to answer these questions by looking in the first instance at the theory of deliberative democracy. In broadest terms, an umpire should seek to consider the issues on their merits and spell out the chains of reasoning that underpin its decisions in terms that are accessible and acceptable to all. It should also urge the parties toward accommodations and fair outcomes. These are, of course, very general formulations. While normatively powerful, they are also practically vague. In order to address this vagueness, and to give substance to the notion of a deliberative umpire, we have also shown how proportionality doctrines already common in related contexts may be used to ‘scaffold’ deliberative agreement in the context of power-sharing vetoes. As noted, more needs to be said about the precise institutional nature of a deliberative umpire and the pros and cons of the different institutional design options. However, by focusing on matters of substance and doctrine, we have sought to fill a key and distinct gap in the theory and practice of power-sharing systems.