A tripartite model of federalism

Abstract

The classical account of federalism is bipartite. Federal systems are understood to have a dual nature: on the one hand, there is the central government, and on the other hand, there are the constituent units. We argue instead for a tripartite model of federalism. In this model, a third institutional pillar is added to federal systems. This third pillar deals exclusively with matters related to the institutional architecture of the system. We argue for tripartite federalism on three grounds: a tripartite structure would be better suited to accommodate political communities where citizens might currently feel misrecognized by the central government, it would provide a more efficient way to adjust the federal architecture and it would be able to do so in a more democratic manner. We conclude our article with a reflection on the distance between our ideal-typical tripartite model and actual reality.

Introduction

Classical accounts of federalism are bipartite. A federal system is understood to consist of two levels of government: the level of the central government and the level of the constituent units. In this article, we argue in favour of a ‘tripartite’ model of federalism. In this model, the two levels are supplemented with an additional, third ‘level’, which should materialize as an institutional pillar in its own right. It is our contention that a tripartite model of federalism offers, in theory, a better interpretation of the logic of federalism. In practice, we believe that a federal system with a tripartite structure would have distinct advantages over a bipartite structure. These practical advantages result from the function of this third institutional pillar: to handle matters of federal-institutional architecture. It would be a forum to discuss alterations of the federal system, to decide upon competence transfers and to adjudicate conflicts between the two main levels of government. In bipartite systems, these tasks are interwoven with the regular institutional life of a federation; in a tripartite system, they are moved to a separate level.

This proposal for a tripartite model is meant as an intervention in two important debates. The first concerns federal safeguards: what are the institutional and non-institutional mechanisms that allow federal systems to maintain a balanced relation between the central government and the constituent units? This question has a long and venerable pedigree. It goes at least back to statements made by James Madison in the Federalist Papers. In No. 51, Madison famously drew an analogy between on the one hand the horizontal separation of powers (Montesquieu’s ‘trias politica’) and on the other hand the vertical separation of powers, that is, the separation between the central government and the constituent units: ‘In the compound republic of America, the power surrendered by the people is first divided between two distinct governments, and then the portion allotted to each subdivided among distinct and separate departments’ (Hamilton, Madison, and Jay 2003: 253). Just as the relation between the departments that are horizontally separated is modulated by a number of checks and balances, so is the relation between the central government and the constituent units stabilized by various such ‘safeguards’ (as the federalism literature calls them today, see e.g. Bednar 2009). For Madison, the most important safeguards were the formal ones provided in the constitution, although he was not blind to the importance of informal safeguards. ¹

These safeguards are necessary because this vertical relationship is fraught with potential tensions. As separate institutional actors, the interests of the centre and of the constituent units do not automatically coincide. Both sides might be tempted, out of sheer opportunism, to push the boundaries or even to transgress them. So there is always the risk of an escalating ‘tug of war’ between central and decentralized forces (cf. Ryan 2011; Erk and Koning 2010; Watts 2008; Bednar 2009). These tensions can have diverse causes. In many cases, it is simply power that is at stake, be it in the form of legislative authority or administrative and executive independence (cf. Riker 1964; Watts 2006). In other cases, financial and economic motives are the dominant factor. When constituent units acquire greater economic relevance, their bargaining power increases which can lead to demands for new fiscal arrangements, for different economic policies, or for readjustments in budgetary flows (cf. Benz and Colino 2011; Brancati 2006; Stepan 1999). In still other cases, claims of identity and cultural (mis)recognition might play a role (Norman 2006). And in many cases, tensions will be caused by a combination of these various incentives. ² These tensions constantly threaten the federal balance of powers and dealing with them is a never-ending task for federal systems (cf. Voigt and Blume 2012; Hueglin 2013). As the literature abundantly shows, federal systems are equipped with a wide variety of safeguards to contain such tensions and maintain the federal balance of forces.

Yet federal systems are also equipped with various devices to alter their institutional architecture. These devices are necessary because federal systems, in order to flourish, need to adapt to changing internal and external circumstances. And the literature shows there to be a wide variety of such devices and mechanisms to institute federal change (Bednar 2013; Benz and Knüpling 2012a; Colino 2013; Hueglin and Fenna 2015). This debate on (constitutional) change in federal systems is the second one to which we wish to contribute.

In the two sections below, we start by looking at the debate on federal safeguards and on federal change, respectively. We conclude the second section by listing a number of problems and weaknesses in currently existent mechanisms to handle ‘federal tensions’. In the third section, we present our proposal for a tripartite model of federalism and in the fourth section we provide an ideal-typical sketch of its concrete institutional shape. In the fifth section, we defend three arguments in favour of tripartism: a tripartite structure is better suited to accommodate political communities that might claim misrecognition by the central government, it provides a more efficient way to adjust the federal architecture, and it is able to do so in a more democratic manner. We conclude with a reflection on the distance between our ideal-typical tripartite model and actual reality.

Maintaining the federal balance

Federal systems typically have several interlocking mechanisms to maintain stability, the most important one being the entrenchment of all federal arrangements in a constitutional text. ³ The distribution of powers between centre and constituent units (who has authority over what matter and whether certain matters are to be jointly decided) is usually stipulated in the constitution, as is the very existence and definition of the constituent units. ⁴ The constitution can also guarantee the participation of the constituent units in the central government ⁵ and typically contains various other norms that are intended to maintain federal stability.

Of course, a constitutional document on its own is powerless; it needs means to enforce itself, starting with a constitutional court that protects the entrenched federal arrangements through substantive judicial review. ⁶ The important role of judicial intervention is usually justified by arguing that the court is objective: ‘its interests are not entangled in the dispute’s resolution’ (Bednar 2009, 120). The ‘ideal court-as-umpire’, Bednar writes, is ‘disinterested’ because it is ‘free of electoral or political influence’ (120–21).

Yet courts are not perfectly neutral. For one, they depend on other powers, for example, the federal government, for the enforcement of their decisions (123). Moreover, judges are always appointed by someone, which can taint their neutrality. A court where half of the judges are appointed by the central unit and half of the judges by the constituent units (as in the German Federal Constitutional Court), will – all other things being equal – be seen as more impartial than a court entirely appointed by the central government (Hueglin and Fenna 2015, 309 ff.). And yet, the overwhelming majority of federal constitutional courts is appointed by their central government, an arrangement – as Hueglin and Fenna note – that ‘is scarcely consistent with the idea of supreme courts as neutral arbiters’ (312). On top of that, a court is always located somewhere. The simple fact that the US Supreme Court is housed in Washington DC strengthens its perceived association with the central government. ⁷ In light of all this, it is not surprising that the historical record shows that constitutional courts have mostly been instruments at the service of centralizing forces. More often than not, they have used the interpretative leeway afforded by the constitution to strengthen ‘a nationally oriented vision’ (312). Thus, it seems that the ‘ideal court-as-umpire’ is at best imperfectly realized in actual federations.

Next to the courts, federations can count on a number of structural and political safeguards (cf. Bednar 2009, 98 ff.). When powers are wisely distributed and shared between two levels of government, these two levels will be dependent on each other and will therefore have to respect each other, but will be sufficiently independent so as to assert themselves when necessary. Wechsler pointed out that the presence of state politicians at the central level (in Congress but also in more informal forums), where they can argue in favour of state rights, is crucial in maintaining a well-balanced relation (Wechsler 1954, 543). Much attention has also been given to the role of political parties. When the different levels of a political party are interdependent, they will – involuntarily – strengthen the federal balance; local party chapters need a powerful ally higher up, whereas national party figures know that they depend on the political achievements of local chapters (Bednar 2009, 113 ff.). ⁸ Yet Bednar also shows the weakness of political parties as safeguards of federalism. Parties are interested in electoral success, not in federal balance per se, so in many scenarios, they might have self-interested incentives to push for centralization (or decentralization) (Bednar 2009, 116-19). Moreover, citizens have relatively little oversight of these mechanisms, leading to questions of democratic accountability. The same goes for other structural or political mechanisms: they are often based on institutional self-interest and on the hope that these interests will somehow keep each other in balance. Yet it is not clear why the people should indeed trust to it that these interests will ultimately work in favour of federal balance.

This brings us to a final and important safeguard of federalism, namely, the people (cf. Bednar 2009, 107 ff.). Because citizens are independent from the two levels of government, they might be able to act as an umpire, so Madison hoped. This presumes that citizens have a ‘cultural sense of balance between the levels’ (Bednar 2014, 243) and that they are sufficiently interested and informed so as to discern the merits of politicians at different policy levels and punish those who threaten the federal balance. Clearly, citizens will not always judge in such a well-informed or neutral manner. As Bednar notes, ‘if citizens identify primarily with one government, then they may forgive or ignore (or even reward) its opportunistic behavior, or be blind to it altogether’ (Bednar 2009, 112). This is true, but it can work for other levels of government at the same time, since in existing federations different groups of citizens tend to identify with different levels of government. Some will identify with the central level of decision-making, but many will identify with both levels. ⁹ Moreover, citizens do not share the institutional interests of politicians. Professional politicians might have a vested interest in maintaining or expanding certain political institutions (because of the opportunities these institutions offer for the advancement of their careers or their party). ¹⁰ In this regard, citizens’ judgement can be expected to be more impartial than that of career politicians.

Evaluating these different types of safeguard, Bednar concludes that ‘there is no such thing as a perfect safeguard’ (Bednar 2014, 244). One could hope for one single safeguard that always protects the best possible interpretation of the constitution, ‘but constitutional truth is subject to interpretation’ (244). Thus, what matters is to bring in a ‘plurality of perspectives’, that is, to bring in views and ideas on the federation from as many angles and corners as possible, so as to come to the best possible aggregate judgement (252). According to Bednar, this plurality is best protected by a variety of safeguards that mutually reinforce each other; their diversity and complementarity making a federal system less vulnerable.

Federal dynamics

Although maintaining the federal balance is of crucial importance, much of the recent literature emphasizes that this balance is a highly dynamic one (cf. Benz and Knüpling 2012a; Benz and Broschek 2013a; Bednar 2014; Resnik 2014). ¹¹ It is now widely accepted that no federal system can flourish without adapting to new circumstances and thus without having the capacity to modify its institutional architecture so as to improve efficiency and accountability (Hueglin and Fenna 2015, 28-31; Colino 2013; Benz 2013). To quote once again Jenna Bednar: ‘a static balance could be the doom of a federal system’ (Bednar 2014, 250).

Furthermore, many authors claim that federal systems today are more dynamic than they used to be. One can point at several causes for this. In the non-Western world, ongoing democratization plays a role (Benz and Knüpling 2012b). But the most important cause is certainly the slow crumbling of the Westphalian order (Calabresi and Bickford 2014, 123-125). States face increasing competition from supranational organizations, and many states are in the process of transferring political and legal competences upwards to these supranational institutions; a paradigmatic case being the new type of ‘coming-together’ federalism embodied by the EU. On the other hand, when states become less important in the global order, this also affects their internal authority and the loyalty of their citizens (cf. Benz and Broschek 2013b). In consequence, there is a growth of subnational claims to autonomy, and many states are today engaged in a slow process of devolution (cf. Keating 2001). At the same time, certain states reassert themselves, try to stop the loss of authority to other levels of government or even actively grab back competences from subnational or supranational institutions; an extreme case being that of the United Kingdom and its attempt to withdraw all competences previously transferred to the EU.

As a result, there is today increased attention for the processes that allow federal systems to change, including – most importantly – the dynamic transfer of competences between different levels of government (cf. Beaud 2014; Watts 2006). If the more traditional debate can be said to deal with the statics of federalism, there is now a growing literature on the dynamics of federalism. Surveying this literature, it is clear that federal change is a polymorphous process, occurring in various places and manners.

As the basic federal architecture is usually constitutionally entrenched, change often needs to happen through formal constitutional amendment procedures. These procedures can resemble ordinary lawmaking but will at the very least be more demanding in terms of the required majorities. Moreover, in federal systems, it is customary that the constituent units play a role in constitutional lawmaking. Constituent units can be represented in a second chamber, but constitutional changes often also need to obtain approval in, say, three quarters of state legislatures, or they have to be ratified in referendums that are organized in every constituent unit separately. The participation of both levels of government in constitutional amendment is so customary that many authors see it as a defining characteristic of federalism. Elazar writes that ‘unless the constituent polities have substantial influence over the formal or informal constitutional amending process, the federal character of the system is open to question’ (Elazar 1987, 173). Ruth Gavison states that ‘it is a central issue of [federal] governments that the division of powers between sub-units and the central government is determined in a way that will be beyond the unilateral change of either states or the central government’ (Gavison 2002, 91-2). Note, though, that this is not everywhere the case. In Belgium ¹² and in Spain, ¹³ neither the executives nor the legislatures of the constituent units have a formal role in constitutional amendment. Inversely, the EU’s central institutions have no official role in treaty changes; the member states’ governments are the sole ‘masters of the treaties’. In general, though, both levels of government are indeed involved in altering federal constitutions, and ideally these procedures show great care in giving similar weight to both levels. This is all the more important as the central level tends to be in the stronger position and as modern states are predisposed to centralize (Levy 2014). In order to find this balance, federal constitutional amendment procedures often involve diverse types of committees and conferences so as to include politicians, experts and/or civil servants from both levels of government. ¹⁴ Again, the underlying ideal is to give a proper hearing to the differing viewpoints of the central government and the constituent units.

Of particular significance are a number of recent attempts to alter federal constitutions through grand, sui generis conventions. There are always local reasons why such a convention is deemed necessary. Yet these conventions also stem from a recognition that in the ‘age of democracy’, the basic structure of society is not something that can be changed in closed doors meetings or through procedures in which only a limited number of formal representatives can participate. Instead, these conventions try to be more inclusive and bring together not only delegates of the two main levels of government but also civil society organizations as well as representatives of other societal groups. They are usually stretched out over longer periods of time in an attempt to also engage the broader public. A modest example is the ‘Austrian Convention’ that was launched in 2003 with the purpose of renewing the Austrian federation by 2005 (Bußjäger 2012). The Austrian Convention was modelled after the European Constitutional Convention, that was held in 2002–2003 and resulted in a first draft for a European Constitution. A very ambitious example is the so-called ‘Canada Round’, a large exercise in constitutional politics that was launched after the failure of the Meech Lake Accord. The Canada Round started with a series of public consultations, which were followed by more formal negotiations between the national government, representatives of the provinces and delegates of indigenous peoples (Verrelli 2012). These negotiations resulted in the so-called Charlottetown Accord which, in a third phase, was rejected in a popular referendum. It is noteworthy that the ‘Austrian Convention’ fared an even worse fate; it never even reached the referendum phase as the negotiating parties failed to find a consensus. The European Constitutional Convention did slightly better. Its draft formed the basis for the Treaty establishing a Constitution for Europe that was eventually signed by all 25 member states, although it failed at the ratification stage after voters in France and the Netherlands rejected the Treaty in referendums. There are case-specific reasons for these failures, but one returning explanation is that these conventions were too ambitious. As one-off events, they were overloaded with a whole gamut of issues and became a target for a wide range of conflicts, ambitions and vexations.

Even though these three conventions failed to produce immediate constitutional change, it would be wrong to write them off completely. The Austrian Convention generated a high degree of public interest and media coverage (Bußjäger 2012, 103), with lasting results. There is now a much broader awareness of the Austrian constitution and its intrinsic flaws, and the Convention’s report has been described as a ‘blueprint for reforms or as a treasure chest full of good ideas’ (Konrath 2012, 333, 347). In Canada, meanwhile, many of the items that were part of the Charlottetown Accord have in the subsequent years been taken up and implemented through various processes other than formal constitutional amendment. ¹⁵ Thus, conventions can be important stepping stones and can be catalysts for ‘a constitutional debate which continues beyond the particular reform process’ (Benz and Knüpling 2012c, 411), even if it is not easy to keep this dialogue going once the convention has passed.

Benz and Knüpling, in the conclusion of their comparative volume on changing federal constitutions, argue that ‘constitutional change in federal or multilevel systems should be regarded as a never-ending task’, and they claim that this ongoing constitutional dialogue should ‘be embedded in the institutional framework of a federal system’ (411). Although they withhold judgement as to what would be the optimal way of doing so, they issue one important guideline:

In order to open the way to opportunities for arguing about constitutional norms and in order to avoid distributive bargaining as far as possible, constitutional policy has to be set apart from normal policy-making and politics. […] What is essential is the definition of a particular arena for constitutional negotiations, the inclusion of actors beyond those dominating intergovernmental and party politics, and to unburden the agenda from issues of normal politics and from party politics as much as possible. (410)

This was one of the reasons why the Austrian Convention failed. The political camps of regular politics were reproduced within the Convention and within its steering committee, with the result that regular party politics took over. According to Benz and Knüpling, this was inevitable. Without a clear distance from ordinary politics, political actors will behave in their regular, bargaining manner. It is only when there is a strong divide that cooperation in a more argumentative style becomes possible (cf. Benz 2013, 84-5).

While formal constitutional change is the most obvious manner to alter a federation’s architecture, other methods exist as well and can be equally important for the dynamism of a federal system. This is especially true when formal amendment procedures are very rigid and constitutional revisions therefore difficult. A case in point is the United States, where, ‘given the difficulty of formal amendment, reinterpretation of the existing rules is the primary means of constitutional adaptation’ (Bednar 2014, 237). The Supreme Court has indeed often put itself in the role, not just of guarding the existent distribution of competences but also of actively altering that distribution through ambitious interpretations of the Constitution. More specifically, the Supreme Court has in the past helped the push for further centralization (Elazar 1987, 214-5; Hueglin and Fenna 2015, 311-2). This can be welcomed as it shows that courts help to provide a federal system with flexibility. But one can also deplore the lack of democratic oversight and argue that the basic architecture of a federal system should be consciously determined by the people themselves (Waldron 2006, 1358). Moreover, even if one does accept the importance of judicial review for federal constitutions, it remains an open question whether courts controlled by the central government (as it is the case for most constitutional courts) can be completely neutral in handling disagreements between the central government and the constituent units. ¹⁶

Courts interpreting the constitutional text in a novel manner are an example of ‘implicit change’. The text of the constitution does not explicitly change, but there is nevertheless a gradual change in the federal balance. There are also other ways in which federal systems can implicitly change (cf. Behnke and Benz 2009, 217 ff.; Kincaid 2012, 38 ff.; Bednar 2013). Parliaments can gradually expand the scope of their jurisdiction, or bureaucracies can start to interpret certain norms in a different way. New practices and institutions can emerge which were not foreseen in the constitution but which do alter the way the constitutional order functions (think of political parties) or which render parts of the constitution dead letter. A federal system can also evolve simply because political discourses in the media offer new perspectives on the appropriate boundaries between different powers and ultimately lead to a new shared understanding of the terms of the federation.

These evolutionary processes are part of the normal life of institutions and are evidence of the constant need for readjustment – a need that cannot be adequately covered by the slow and exceptional procedures of constitutional politics (cf. Kincaid 2012). ¹⁷ But there are of course shortcomings to implicit change. The federation remains stuck to the text of the constitution; detailed written constitutions in particular are unlikely to undergo major changes through implicit methods. And again, one could complain about a potential lack of democratic oversight. Implicit change can mean creeping centralization (or creeping decentralization). From the perspective of democratic legitimacy, giving the people conscious control over the institutional architecture of their federal system seems preferable.

The lesson to be learnt from this brief walk through the literature is that the federal balance is always a highly dynamic one, with permanent readjustments. In other words, there is a ‘Madisonian’ need for stability, as well as a ‘Jeffersonian’ need to leave sufficient room for alterations and renewal. Assessing the current mechanisms to satisfy these two needs, several weaknesses and concerns appeared.

Federal conflicts need to be adjudicated by a neutral arbiter, but it is not clear where this neutrality can be found. Even constitutional courts are not necessarily neutral, as they tend to be controlled by central governments. Nor will political or structural safeguards automatically result in a neutral solution; these safeguards are based on the interplay of groups pursuing their own institutional interests, and there is no reason to assume that the resulting balance will always be the optimal one for the federal system as a whole.

In making adjustments to the federal architecture, it is important that a plurality of perspectives is taken into account. In particular, both levels of government should be involved in the process and receive similar weight, which is especially important given potential centralizing tendencies. But it is also imaginable and probably preferable to include an even wider range of perspectives and actors; such as representatives of various societal groups, as has been attempted in recent conventions. Most existent procedures do not allow for such an inclusive modus operandi.

The Canada Round also included public consultations, which suggests that, in the age of democracy, it might be appropriate to give citizens a clear role in the process (and not just at the very end, as in ratifications by referendum). This is especially relevant because the citizens are neutral in at least one regard: they lack the vested institutional interests of career politicians. Yet most existent procedures have not been shaped with a strong concern for democratic participation in mind.

At the same time, constitutional conventions, as one-off events, are not the ideal format to sustain an ongoing constitutional dialogue. As pointed out by Benz and Knüpling, it would be better to firmly embed such a dialogue in the institutional framework of a federal system. Ideally, this would be done in such a way that this dialogue is shielded from the jumble of daily politics.

Our proposal for a tripartite model of federalism is meant to show the direction of a potential solution. We believe that the creation of a permanent forum that is dedicated to deal with all meta-institutional matters could remedy some or even all of the weaknesses mentioned.

A tripartite proposal

We accept the premise that federal systems are intrinsically dynamic and that this will be even more true in the future: federal systems will face more destabilizing challenges than they currently do, competences will be increasingly mobile (i.e. there will be more competence transfers between different levels of government) and there will be more calls for institutional change. In light of this, we defend the following claim. The logically purest federal way to alter the federal architecture consists in processing such alterations not within the orbit of the central unit, nor in the circle of the constituent units, but in a third place where the two levels of government can meet without any presumption of hierarchy between them. The more this forum is independent from the other two levels, the more successful it will be in maintaining the federal balance of powers and in making adjustments where necessary.

This is certainly not a novel idea. John Stuart Mill already argued that competence disputes should be settled by a distinct ‘umpire’:

Under the more perfect mode of federation, where every citizen of each particular state owes obedience to two governments, that of his own state and that of the federation, it is evidently necessary not only that the constitutional limits of the authority of each should be precisely and clearly defined, but that the power to decide between them in any case of dispute should not reside in either of the governments, or in any functionary subject to it, but in an umpire independent of both. (Mill 2015, 381)

Mill was primarily thinking of the adjudication of conflicts by courts, but we believe the same holds true for political processes. Political debates and decisions concerning matters of federal design would ideally be located in a forum that is as independent as possible from the existent political processes. This does not mean that this forum should be entirely disconnected, but it should be perceived as being under the sway of neither the constituent units, nor the central unit. As this forum is the place where all ‘federal tensions’ are being dealt with and where binding decisions concerning the federal architecture, including the allocation of competences, are made, this forum effectively becomes the site of the Kompetenz-Kompetenz. And this is precisely the point of our tripartite model: in our view, in a federal setting both the juridical Kompetenz-Kompetenz and the political Kompetenz-Kompetenz should be evacuated to a space that is separate from the regular centres of authority.

In what follows, we will defend tripartism as the appropriate model for federalism. By ‘model’, we mean three things: interpretation, reconstruction and prescription.

First, tripartite federalism is an interpretive model that wants to capture the intrinsic logic of federalism. Federalism seems to contain a specific promise: the promise that authority will be divided between two levels of decision-making that stand on equal footing. This division of authority needs to be entrenched in some way. If the central government can change the division of authority at will or freely redraw the boundaries between the various lower entities, then, as far as the division of authority is concerned, there would be no difference between federalism and a decentralized unitary state (cf. Rubin and Feeley 1994, 910 ff.; Hueglin 2013, 44). Similarly, if the constituent units could at will withdraw all competences from the central government, federalism would become indistinguishable from confederalism. Thus, the federal ideal of balance seems to require that authority conflicts (or, more generally, decisions about the structure of the federal system) are not settled exclusively by either of the two levels of government. There has to be a capacity to make institutional decisions and alterations. But if this formal capacity is situated at either the level of the central unit or at the level of the constituent units, a federal system does not live up to its promise and is actually tilting towards unitarianism or towards confederalism. The purest, ‘ideal-typical’ way to escape the conceptual pull from unitarianism on the one side and confederalism on the other is to vest the power to alter the federal system in a third, separate entity so that any relation of hierarchy between the different policy levels is avoided. In other words, we claim that a tripartite structure offers the best institutional translation of the idea of federalism.

Second, our tripartite model is meant to be reconstructive. Although there are no examples of federal systems with a fully fledged set of separate institutions dedicated to questions of federal architecture, we believe that a tripartite logic is implicitly at work in many existent federations. These federations might see themselves as bipartite, but bits and pieces of their procedures actually point to a tripartite modus operandi. This article is written from the perspective of normative political theory, so we cannot offer an empirical survey of federal systems and the various tripartite elements they might contain. Nor can we fully assess whether tripartite mechanisms do indeed always work better. Nevertheless, we will claim that our model, rather than a mere piece of political fiction, offers a systematic reconstruction of a logic that is already used, with varying degrees of success, by certain contemporary federal systems.

Thirdly, our tripartite model is a prescriptive model. It is a normative theory that hopes to offer guidelines or at least inspiration for institutional change. Of course, we do not claim that all bipartite federal systems are somehow misguided and should be turned into tripartite systems. In our view, federalism well understood would find its most complete institutional expression in a threefold political architecture. But this is not a reason to tinker with well-functioning bipartite systems. (At most, it is a reason to call them slightly biased towards unitarianism or towards confederalism.) However, federal systems that are burdened with deep, permanent authority conflicts – and, as a result of this, with a lack of political efficacy and legitimacy – might do well to think through the nature of federalism and consider institutional changes that would make them more tripartite. Such a structure would realize, more fully than a bipartite structure can, the federal promise of symmetry between the central unit and the constituent units and might therefore help to handle federal tensions in a fair and democratic manner.

According to our argument, the implementation of this third forum turns bipartite into tripartite systems because it gives material substance to a logically necessary feature of a federal system. Federalism divides political authority over several levels. But to do so, some instance must deal with the question who gets to hold what portion of authority (and what to do in cases of authority conflicts). We suggest to dedicate this question to a third level of the federation. This makes the system tripartite because there are now three distinct parts: the organs of the central unit, the organs of the sub-state units and the organs of the third pillar which deals with the Kompetenz-Kompetenz. Let us clarify this point further.

In a federal state, the central government’s authority is only partial, just like the component units only have partial authority. This is the result of the federal partitioning of political authority. Because of this partitioning, the central government cannot embody the whole of the authority in a country. In our tripartite proposal, the real whole consists of the three levels taken together. Thus, the purpose of this third entity we propose is not to embody the whole. Instead, this third level is what remains of the ‘whole’ authority of the state excluding the central and the component units. The power of this third level, then, is very limited: it tackles those matters that concern the relationship between the other parts, since all other matters are the prerogative of these parts themselves. It is possible to attribute this third task to the central level, but then one undoes the system’s federal logic and makes it more unitary. ¹⁸ Inversely, one can attribute this task to the component units, but this makes the system akin to confederalism.

One good reason for exploring the potential of a tripartite model, is that this has hardly been done before. Most of the literature seems firmly wedded to a bipartite interpretation of federalism. However, important exceptions are, on the one hand, the French theorist Olivier Beaud, as well as – much earlier – the German jurists Albert Hänel and Hans Kelsen and, on the other hand, the German scholar Uwe Leonardy. ¹⁹ In his Théorie de la Fédération, Beaud (2007, 133–43) explicitly advocates a tripartite theory of federalism, as opposed to the traditional, dual understanding of federalism. Yet Beaud does so in order to remedy a definitional problem. Beaud wishes to make a clear semantic distinction between the central government and the entire federal system. He therefore proposes the term Fédération (with capital F) to denote the whole of the federal system, that is, the state-wide unit and all the sub-state units taken together. Fédération is an abstract, analytic term introduced to maintain terminological neutrality: we should not label the whole with the name of one part of it. Thus, Beaud’s core purpose is to clearly distinguish the federal system in its entirety from the state-wide unit. Our project is a very different one. In our proposal, the third part does have a real, institutional life of its own. It is no longer an abstract term to think the whole but has specific institutional content apart from the other two levels.

Beaud is inspired in this by the older work of Hänel and Kelsen, who similarly advocated the usage of a separate word to indicate the federal system as a whole. Hänel used the term Bundesstaat, Kelsen (in his book’s English translation supervised by himself) used the term federal state. For both of them, this term comprises the central governing unit (for Hänel the Gesammtstaat; for Kelsen the federation) and the constituent entities (for Hänel the Einzelstaten; for Kelsen the component states) (Hänel 1873; Kelsen 1949). In all these cases, the usage of a tripartite model is a matter of terminology, introduced to achieve greater conceptual clarity in theoretical discussions on federalism and to give us a new way to conceptualize a federation. Thus, these tripartite models are interpretive at most. Beaud, Hänel and Kelsen object to the common mistake of silently identifying the central level with the whole of the federal system and introduce a third term to correct this. But they do not propose to erect new institutions. As Kelsen noted (1966: 262): ‘the organs of the federation are at the same time organs of the total community, the federal state. For this reason, the federation, as a partial community, is usually not distinguished from the federal state, as the total community’.

It follows that their ‘tripartism’ has no practical consequences: the same ‘organs’ remain in existence. The third part of the federal tripartism of Kelsen and Beaud does not live outside the confines of theory; it is simply a term introduced to denote the whole. One common criticism, then, is that their theory is fictional; in this vein, for instance, Anna Gamper argues that Kelsen’s third level is ‘usually unknown to federal states’ (Gamper 2005: 1304). Our proposal is markedly different because it holds that the third part of federal tripartism should be embodied in a really existent institutional body.

Moreover, the extensional content of our third part differs from what Kelsen and Beaud call the third part. Their third part denotes the whole. Our third part does not denote the whole, but really a third part. If someone wanted to use a term to denote the whole of the threefold system we are proposing, a fourth term would have to be invented. If we applied the conceptual apparatus of Kelsen and others to our proposal, our proposal would become quadripartite, as there would now be four elements: the central unit, the sub-state units, our third pillar and the sum total of the previous three.

Apart from Kelsen, Hänel and Beaud, there is also the work of the federal theorist Uwe Leonardy, who goes one step further and defends something significantly closer to our proposal. Analysing Germany’s federal system, he distinguishes two types of intergovernmental institutions that deal with the relationship between the Federation (Bund) and the Länder: the ‘Whole State’ (the Gesammtstaat) and the ‘Federal State’ (the Bundesstaat). While the former operates on the basis of compromise, the Federal State works through majority-voting (Leonardy 2012: 320-322). Leonardy sees both types at work in Germany. The key institution for the Whole State is the four-monthly ‘Conference of the Heads of Governments of the Federation and the Länder’ (320). The key institution for the Federal State is the Bundesrat (321). What Leonardy does is to theoretically project the third level of our tripartite proposal (which he breaks up into two types) onto Germany’s existing institutions. He theoretically reconstructs German federalism as employing a logic that transcends bipartite federalism.

While we concur with this reconstructive work (as per the second function of our tripartite proposal), our model is, in addition, prescriptive, and we think the tripartite logic is to be institutionally implemented and recognized as such. We argue that federal systems should in practice erect a really existent ‘third’ level where issues that concern the institutional architecture of the federal system can be discussed, decided and adjudicated.

Tripartite federalism in practice

The forum that we are proposing could take various forms, depending on the type and history of the existent political structures. In consequence, its institutional lay-out cannot be completely determined a priori. However, for the sake of clarity and purpose, let us briefly sketch an ideal-typical image of how this forum or ‘third pillar’ could work in practice.

The policy scope of this forum is very limited: it is dedicated to resolving tensions between the central unit and the constituent units and making decisions concerning the institutional architecture of the federal system. In spite of its narrow focus, this third entity would have legislative, executive and judiciary tasks and hence needs the corresponding institutional organs. The legislative assembly has the important task of deciding upon institutional questions. Hence, a sizeable assembly – or even a bicameral set-up – seems advisable. In contrast, the executive organ, which would oversee the implementation of these decisions, can probably be limited in size; it will never have the capacity to coerce the other governments into compliance so it inevitably depends on collaborating with them. The fair and neutral adjudication of federal conflicts, in particular conflicts concerning the distribution of competences, calls for a specialized court. Although comparable to existent constitutional courts, our proposed court would have a more credible claim to independence and impartiality as it would not be associated with either the central or the constituent units. Precisely because all these institutions should have a high degree of independence, it might be advisable to establish their premises in a distinct city. In the case of the EU, one could, for instance, repurpose Strasbourg and make it the official seat for the new, third pillar. The spatial distance (in this case, the distance from Brussels and from the other capitals) would help to mentally disconnect this third pillar from a federal system’s other institutions.

The crucial component in this pillar is clearly the legislative branch, which is to make contentious decisions about the federal architecture in a visible, democratic and efficient manner. There are good reasons to be pessimistic about its potential to do so. The legislative assembly has to couple the horizontal logic of a democratic arena, which is normally determined by ideological divisions and public deliberations, to the vertical dynamics of multilevel negotiations and coordination, an arena that is typically dominated by the executive branches. As Arthur Benz has recently stressed, this coupling exercise is what federal democracy is all about: the institutional conditions of democracy need to be combined with the mechanisms of intergovernmental interactions (Benz 2021). This combination is particularly challenging in the sensitive domain of federal reform, where the powers of all those involved are at stake. More concretely, the composition of the legislative branch will have to navigate the following obstacles.

The process of constitutional change in a federal system is often controlled by the executives, with legislative assemblies or the citizenry only coming in to ratify the results. This executive dominance is problematic from a procedural perspective, but also in light of the outcome as it allows them to maintain and entrench their own position (Fossum and Laycock 2021, 74-75). A prime example is the EU’s national governments, who consistently preserved their own status as ‘masters of the treaties’ through numerous rounds of reform. In order to include but also restrain the executives, a bicameral set-up might be considered, with a second chamber reserved for government members (of the constituent units and of the central unit). The inclusion of the executives is necessary, among other reasons, to keep them responsible. After all, it is these governments that will have to operate within the resulting institutional structures; if they had no say in shaping these institutions, they might deflect responsibility for inefficient outcomes. At the same time, it is crucial to find mechanisms to restrain their power.

As to the first chamber, it seems wise to mix three types of representatives: (i) members of the parliaments of the constituent units, (ii) members of the central parliament and (iii) directly elected representatives. The third pillar needs to be independent, but it also needs connections to the public sphere and the democratic life of the constituent units and the central unit. A completely disconnected forum will not retain the citizens’ attention and will not generate democratic legitimacy. Moreover, it is important that all relevant concerns and views are brought together in this forum. Members of the central and the constituent parliaments are probably best suited to establish permanent channels of communication. An embryonic example is provided by the various interparliamentary conferences of the EU, where selected members of the national parliaments convene with selected members of the European Parliament. ²⁰ This example already reveals some of the practical difficulties. For instance, it is not obvious that the members of the central parliament recognize the other parliamentarians as their equals: MEPs tend to insist on a special status, for instance, by having a larger delegation than the other parliaments or by permanently co-chairing its meetings (Cooper 2016; Crum 2020). It has also been observed that the relations between the different parliaments tend to be rivalrous and competitive, already at the stage of setting the rules but also in the further life of these interparliamentary conferences (Kreilinger 2018). Moreover, the conflicts in these interparliamentary contexts are ever less defined by ‘national or party-political cleavages (…) and ever more by parliamentary levels, sometimes even dividing MEPs and MPs belonging to the same country and political party’ (Herranz-Surrallés 2014, 960). This competitive dynamics is most pronounced in areas where both levels want to claim ultimate authority (ibid.). Those hoping for peaceable interparliamentary cooperation in the EU will find this situation regrettable. And sometimes these conflicts might indeed prove intractable. But to the extent that this competition for authority reflects underlying societal divisions, the presence of a forum where these divisions can be visibly expressed and explicitly resolved might well be an advantage. In order to attenuate these conflicts, but also to enhance the democratic character of the tripartite assembly, it seems recommendable to include directly elected representatives, who would campaign specifically around institutional questions. ²¹ These representatives would have no formal ties to the other parliaments and would increase the citizens’ democratic control over the federal architecture. Electing these representatives in a separate election, at a different time and on a different rhythm from other elections, would again help making this legislative recognizable as an independent institution. The legislature would of course have to be structured in such a way as to prevent the dominance of any one group over the others: by carefully balancing the number of representatives, by setting up special majority requirements and maybe even by giving veto rights to certain constellations of representatives.

An important stumbling block for honest deliberation in this legislative assembly could be the intrusion of party politics. ²² As was illustrated by the Austrian Convention, the partisan divisions from regular politics can contaminate and spoil the domain of constitutional debate. The spoiling effect of party politics in federal intergovernmental relations is of course strongest when the conditions for party incongruence are fulfilled, that is, when some or all of the same parties operate in different polities and at different policy levels (Bolleyer, Swenden, and McEwen 2014). These parties can function as communication channels and hence have important positive effects (Benz and Sonnicksen 2021, 257–258). But in periods of incongruence, when opposing parties govern at different policy levels, their competition might inflate the divisions between these policy levels and prevent consensus-seeking behaviour in the federal reform process. Proper attention should also be given to the specificity of the local party landscape. With regard to the regular political life of federations it has been demonstrated that party competition can be a boon to a healthy federal dynamics but can just as well be a threat to it (Detterbeck, Renzsch, and Kincaid 2016); undoubtedly, the same goes for the politics of federal reform. ²³

There are further institutional elements that have to be taken into consideration when designing a successful tripartite assembly. For instance, it has been reported that interparliamentary activism is stronger where the legislative branch has a high degree of independence (as is illustrated by the National Conference of State Legislatures in the USA) and less strong in parliamentary systems, where the executive and the legislative majority are fused (Bolleyer 2017). Members of parliament might also have various other reasons for being reluctant ‘multi-arena’ players, as has been demonstrated in the context of the EU (Auel and Neuhold 2017). Another consideration is the structure and the division of the work packages. In comparing recent processes of federal reform in Germany, Austria and Switzerland, it has been noted that the Swiss approach, where the general principles were settled in a first phase and the concrete details and negotiations were reserved for a second phase, offers distinct advantages (Behnke 2010, 44).

It also seems advisable for the legislative chamber to host hearings with civil society organizations and other relevant groups, groups that might have no constitutional standing (and are therefore often not formally heard in currently existent procedures) but whose input should be taken into account when major institutional decisions are prepared. As Benz has noted, the inclusion of these groups ‘can contribute to balancing the impact of party politics and to moderating confrontation or strategic bargaining in intergovernmental politics’ (Benz 2021, 54; cf. Detterbeck, Renzsch, and Kincaid 2016).

It is also recommendable to give a sufficiently strong role to regular citizens. This is in line with the recent trend towards more inclusive and democratic forms of constitutional politics (cf. Colón-Ríos 2012; Zurn 2016). The inclusion of non-professional politicians not only increases the democratic legitimacy of the reform process but can also provide another counterweight against partisan calculations. As is already the case for constitutional amendments in numerous countries, all major decisions can be linked to a referendum. Of course, referendums come with their own risks, and we agree with much of the contemporary literature that referendums should always be embedded in a broader, deliberative system (Tierney 2012; Parkinson 2020). That is why it might be wise to aim for a more complex set-up, for instance, comprising citizen panels drawn by lot mixed with professional politicians, with added feedback through new and old media and a referendum as its closing piece. These were the ingredients of the recent Icelandic experiment in ‘inclusive constitution-making’ (cf. Landemore 2015). Although not in a federal context, the Icelandic concerns were clearly the same as ours, namely, how to provide society with the capacity to collectively reflect on fundamental institutional issues in an impartial manner and how to give reform processes a high degree of democratic legitimacy. We are encouraged by findings, such as those of Nenad Stojanovic (2006; 2011), which confirm that even in a complex set-up with multiple sub-units and sub-national groups, such as Switzerland, forms of direct democracy can have centripetal effects and do not necessarily exacerbate conflicts.

The most important ingredient for the success of this tripartite federal reform process is probably the one that is least susceptible to institutional engineering, namely, the presence of the right ‘spirit’. Citizens and politicians need to trust and find each other in a deliberative and cooperative mood in this tripartite setting, at a safe distance from the bickering of daily politics. When successful, this would give citizens the feeling that their federal institutions, rather than a fragile truce or a mere modus vivendi, are indeed reflecting their multilayered society and are the outcome of their own ideas and preferences. Creating such a ‘federal culture’ is maybe the most significant task of the tripartite institutions, but also the most difficult one.

It should be stressed that our argument does not assume that all federations should be engaged in a permanent process of reconfiguration. Institutions and policies shape the plans, expectations and even the identities of citizens. Therefore, a considerable degree of stability is desirable. At the same time, flexibility is needed in light of new circumstances and sensitivities. In order to strike a balance between the need for stability and the need for change, the activities of this new legislative forum can best be tied to a relatively slow working cycle, for instance, a cycle of 10 years. This implies that a new series of talks is launched every decade, with probably greater public interest at the beginning and at the end of every cycle. It is not even necessary for this forum to convene on a continuous basis. In very stable federal systems, it might only gather when the need arises; this would also prevent overzealous representatives from artificially increasing intrafederal tensions so as to justify their own position.

These and other institutional specifics are important, but we can only discuss them in a cursory manner here. And, of course, none of these practical considerations is sufficient to guarantee that our tripartite proposal would indeed outperform existent procedures for federal reform. Rather than providing a complete blueprint, we simply wish to nudge the reader’s imagination in a specific direction: would it not be more in keeping with the spirit of federalism if the ongoing constitutional dialogue that many federal systems need, would be organized, not by the central government, nor by the constituent units, but by a third set of ‘thin’ institutions that stand at a safe distance from these powerful players? Our main purpose in this article is to establish the argument that federal systems should ideally have a separate forum that is not in the thrall of either the central unit or the constituent units and that is exclusively dedicated to handling intra-federal tensions. The institutional specifics can and should differ from federation to federation.

Three arguments in favour of tripartite federalism

We believe that a tripartite model, along the lines we just sketched, would be able to remedy some of the weaknesses of currently existent procedures for federal adjustment. More specifically, we believe that the choice for a tripartite structure would be justified in light of the following three considerations.

Argument 1: Recognition

This first argument is especially applicable in federal systems where the rationale for federalism is the accommodation of national-cultural groups. Resnick and Kymlicka distinguish between such recognition-related federal systems, which they call ‘multination federalism’, and ‘territorial federalism’, where federalism is not related to the recognition of cultural identities (Resnick 1994, 71; Kymlicka 2001, 91–119). In multinational federations, such as Canada, Ethiopia, Belgium or Spain, citizens and politicians identifying with the constituent nations often look with suspicion at the central government and want to ensure that their constituent unit is not treated as subordinate. In this context, it is particularly important that decisions about the federal architecture are not dominated (or perceived to be dominated) by the central government. Locating these decisions in a third pillar would help in preventing the central government from being seen as the hierarchically dominant unit.

It would also guarantee that changes to the institutional framework are made in an ‘explicit’ way, that is, within carefully balanced procedures where all concerns can be publicly voiced and acknowledged. In the sensitive context of multinational federations, implicit change (for instance, in the form of creeping centralization) is certainly not innocent and could be a source of increased tensions. Clearly delineated procedures, which give citizens in constituent units the feeling that they are respected as equal discussion partners and not subordinated to another level of government, are preferable in this context.

Yet the concern for recognition is not only relevant when cultural identities are at play. This same concern is tangible, for instance, in the American debate on ‘dual federalism’ as it was recently revived by Sotirios Barber (Barber 2014). Dual federalism implies that the central government and the constituent states are levels of government that stand on equal footing and that are equipped with separate and mutually exclusive spheres of action, spheres that should be carefully protected against possible incursions by the other side. ²⁴ According to Barber, dual federalism is a self-defeating position. Those who defend the rights of states because they feel that the national government has violated them, inevitably do so in a national forum. However, by submitting ‘state-federal conflicts solely to the nation’s agents’ (Barber 2014, 6), they already acknowledge that the two levels do not stand on equal footing. As soon as they admit that the ‘states’ rights debate is a national debate, conducted in a national forum’, they practically accept that the states depend on the nation. Thus, according to Barber, the defender of dual federalism ends up in a performative contradiction. She submits her defence to a ‘national judge of some sort, like Congress or the Supreme Court’, while in principle denying that there is an authoritative national judge (Barber 2014, 5).

Barber’s position has been vehemently criticized. Logically, the sheer fact of speaking in a national forum does not commit a speaker to accepting the authority of that forum; one can intervene simply to make it clear that the national context is not the appropriate forum for the debate and that one has no further intention of participating in the debate there (Blake 2014, 23 ff.). ²⁵ Descriptively, Barber’s point is even more difficult to defend. It is an undeniable fact that in constitutional courts and national parliaments of federal systems, many actors do plausibly argue for the limitation of central authority (Young 2014, 62). But we think that Barber’s point is valid in one important sense, namely, by taking into account the symbolic effects of institutions. When representatives of constituent units are forced to present their claims and arguments in a forum that clearly belongs to the national context, they are indeed – at least symbolically – recognizing this context as authoritative. And it is doubtful that, within this context, representatives of the constituent units and appointees of the central government will ever be able to genuinely recognize each other as equals. Moreover, there is in every forum a certain ‘esprit de corps’. This is clearly visible in the European Parliament. In principle, it is possible to be a member of the European Parliament (MEP) and use this venue exclusively to argue that it should be abolished (think of the former delegates of the UK Independence Party). In practice, however, MEPs with an outspoken anti-European attitude inevitably run against the collegial atmosphere that characterizes most debates in the Parliament. Thus, if the ideal of federalism encompasses the value of equal recognition between two levels of government, it seems wiser to move debates where the interests and views of these two levels are at odds, to an independent forum. Otherwise, constituent units that already feel dominated by the central government might have the impression that in the very act of voicing their complaints they are forced to genuflect once again for this government.

Argument 2: Efficiency

Above, we already mentioned that the extraordinary procedures of constitutional politics can be spoiled by spill-over effects from ordinary politics. Benz and Knüpling, for instance, use this as an argument to plead for a strong separation between these two types of politics. Yet the inverse is also true: disagreements about the federal architecture can threaten the efficiency of regular policy-making. When fundamental institutional questions come to interfere with the business of daily politics, be it at the level of the central government or at the level of the constituent units, the mixture may lead to instability, and in extreme cases, it can even force regular politics to a grinding halt.

A clear case in point is Belgium, a federal state with significant distrust between the different component communities. Alterations to Belgium’s federal constitution are made exclusively at the central level, with no formal role for the constituent units. ²⁶ An outgoing government can ‘unlock’ specific constitutional articles, while the new government, composed after intervening elections, can make actual alterations to these articles. In consequence, the end of a legislature and the formation of a new government is often an opportunity for debates and decisions about constitutional change. However, if the parties that are about to enter the coalition government do not find agreement on such changes, the whole process of coalition formation is in peril. At occasions, this has made federal government in Belgium close to impossible. During the longest cabinet formation period in Belgian history, Belgium was without a new federal government for 541 days (from June 13, 2010 to December 6, 2011; see Devos and Sinardet 2012). During this time, parties did not manage to form a new coalition, not because they disagreed over substantial policy issues (there surely was such disagreement, but not of an unusual nature) but because the dominant Flemish parties wanted further decentralization, whereas Francophone parties were unanimously opposed to further decentralization. The cause of this gridlock was clearly the mixture of functions at Belgium’s central level: the central government has a whole gamut of regular policy tasks, but it is also a forum for decisions about the federal structure and about the distribution of policy tasks between the two levels of that structure. The Belgian example shows that folding these two functions into one and the same institution can be highly paralysing.

The efficiency argument has resonance beyond Belgium. At the height of the financial crisis the Bundesverfassungsgericht, Germany’s constitutional court, claimed that the ‘Outright Monetary Transactions’ (OMT) were a transgression of the European Central Bank’s (ECB) legitimate scope of action and asked the Court of Justice of the European Union (CJEU) for a preliminary ruling about the matter. ²⁷ Under the OMT-programme, the ECB purchased sovereign bonds of threatened Euro-members on secondary markets, in view of stabilizing their price. According to the ECB, this programme was a vital technical tool without which its regular operations (creating price stability through varying interest rates) became impossible. According to the German court, however, the OMT-programme was a piece of economic policy that resulted in fiscal redistribution among member states and hence exceeded the ECB’s competences. This case shows that disagreement about competence domains can burst into the open at inconvenient times and potentially disrupt urgent policy actions. This risk can never be completely eliminated, but the current way of proceeding is dysfunctional. If disagreement about the delineation of competences expresses itself in such a potentially debilitating way (one member state’s constitutional court nearly disabling the policy instrument on which, according to the ECB, the EU’s very fate depended), it is because this disagreement does not find a proper outlet within the EU’s bipartite set-up. In the current situation, the CJEU and the Bundesverfassungsgericht, two courts that are perceived as partial (respectively in favour and critical of further EU-centralization), fight a proxy war over what is in fact an existential political question. ²⁸ In a tripartite set-up, a neutral court and a democratic public forum would be available to house these disputes.

The absence of such a forum in the EU has further pernicious effects. Several Eurosceptic parties believe that too many competences have already been transferred to the European Union. It seems only logical that this point of view is represented and receives a fair hearing. But as things currently stand, the only Europe-wide stage available to these parties is the European Parliament. In consequence, they regularly use that forum to question the concentration of competences at EU-level, even if such interventions are out of place there given that the European Parliament is only designed to deal with the substance of European policies and has no real say in institutional questions. Other parties have turned to similar tactics of obstruction. For example, several far-left groups (such as the Bulgarian SKB, Romania’s PP-DD, the Dutch SP or Slovenia’s Združena levica) have abandoned the strategy of steering the EU towards a more left-wing course and see little reason to further cooperate in policy-making in European institutions. Instead they direct their political energy at bringing competences ‘home’ (where they hope to find more left-leaning electorates) and sometimes at simply obstructing EU policy-making. Yet again we see here an instance where day-to-day policymaking is impeded by competence- questions. If we could build an institutional wall between both discussions, competences could be exercised more efficiently.

As federal systems become increasingly dynamic, we can expect such scenarios to become more widespread in the future. Admittedly, even in a tripartite system, parties in favour of constitutional change could hinder regular policymaking at the central level. But at least it would be obvious to the public eye that they are directing their energy in the wrong direction and that they should rather make themselves heard in the federal system’s third pillar. In very conflictual federations, the slow working cycle we propose for this third pillar (e.g. every decade) would also have a stabilizing effect as it would be futile to permanently campaign for institutional adjustments.

In summary, the efficiency argument reads that intense disagreements over the future of a federal order should not impair ongoing decision-making processes in regular political forums. We submit that this risk can best be averted by establishing clearly separate places for daily politics and for constitutional politics.

Argument 3: Democracy

A third argument in favour of tripartite federalism is the normative requirement of democratic legitimacy. Existent bipartite federal systems have various mechanisms in place for institutional adjustment but, as mentioned above, they are not always good at providing genuine democratic oversight over these adjustments. There at least two reasons why popular control could be improved by dealing with institutional matters in a separate forum.

First, there is always a risk that the relevant interests and forces that are at play in authority conflicts, are not adequately revealed to the citizens. When it comes to transferring a competence from its current location to another decision-making level, at least three different sets of interests can be distinguished: the interests of all affected citizens, the interests of representatives and personnel of the decision-making level currently exercising this competence and the interests of representatives and personnel of the decision-making level to which the competence is potentially transferred. In the EU, for instance, these different interests are clearly not adequately represented. Transfers typically happen through new treaties (or treaty changes), which are negotiated between the member state governments, that is, by politicians representing only the lower policy level. In many cases, these politicians can be expected to have an interest in keeping the competence at that policy level, regardless of whether this serves the general interest. ²⁹ Inversely, putting these decisions in the hands of the central government would be, as some have described it, like ‘leaving the foxes in charge of the henhouse’ (Young 2014, 47). This danger is particularly real because a central government is often ‘able to present itself as neutral’ in comparison to the more ‘‘particular’ wills vested in smaller political entities’ (Weinstock 2014, 264). For instance, central governments tend to see themselves as the most suited interpreter of the principle of subsidiarity. Yet there is no reason why central governments would always be neutral; at least in some cases, its appointees will be biased in favour of centralization. It is therefore of crucial importance that all categories of interests are properly represented and made visible in the decision-making process, so that this process is not skewed in favour of either the central government or the constituent units. And it is crucial that the citizens themselves, who are at least neutral in the sense that they generally lack the vested institutional interests of professional politicians, have a strong say in the ultimate decision.

A second weakness in many existent mechanisms for constitutional adjustment, results from the fact that political and electoral procedures in bipartite systems were not designed for the purpose of federal institutional conflicts. They were rather designed for ‘regular’ political conflicts, conflicts that are typically structured around left-right (or progressive-conservative) oppositions and that do not always map onto disagreements about the federal architecture. The latter disagreements often revolve around considerations of subsidiarity and efficiency, and – as is well known – subsidiarity as a principle is not a neutral guide but always requires substantial interpretation (cf. Bermann 1994; Føllesdal 1998). Moreover, subsidiarity is not a controlling principle, but is simply one possible rule that self-determining polities can adopt to think about the distribution of authority (Weinstock 2014, 261-3). Distributing competences can also require a judgement about feelings of national or subnational identity and about the relevance of such feelings for the allocation of competences. Moreover, different groups can have radically different conceptions of the type of federalism most suited for their country. ³⁰ The lack of congruence between all these considerations on the one hand, and the classical ideological positions of domestic political conflict on the other, makes existent democratic procedures rather unsuited to express, channel and resolve disagreement over federal institutions. ³¹

In short, in absence of specific procedures dedicated to institutional matters, it is unlikely that all relevant ideas and interests are properly taken into account, be it because the decision-makers have a vested interest in one specific outcome or because institutional questions are forced into the ideological shapes of daily politics. This puts into question the democratic legitimacy of the resulting decisions, even if all relevant procedures have been followed. On a strong conception of democracy, democratic legitimacy requires the presence of a stage, visible to all citizens, where all relevant policy issues are laid out and where the tensions between the different forces, interests and principles are dramatically brought to the fore (cf. Lefort 1988; Føllesdal and Hix 2006; Hix 2008; Rummens 2012). Only in this way can there be a proper public debate, can citizens become aware of the different options that are available and can they consciously recognize the resulting decision as their own decision. This is what might be missing in many existent procedures: the stakes and principles are often insufficiently visible and hence poorly understood by citizens. Setting up a separate entity that is primarily concerned with institutional matters, might solve this problem. Disconnected from the business of regular politics, this would be a forum where the interests and principles that are at play in alterations to the federal architecture could be discussed in their own terms and properly brought to the attention of citizens.

Setting up such a dedicated forum would also be the most straightforward way to bring a large multitude of perspectives to bear on questions of institutional design so as to come to optimal decisions. ³² Arguably, such a multitude of perspectives can also be mobilized by having a wide range of different mechanisms for institutional change, mechanisms that are spread out through the federal system and that are ‘interlocking’ or keep each other in check (this seems to be Jenna Bednar’s preferred option). ³³ However, in keeping with the literature on deliberative democracy, we believe that bringing together all relevant interests and considerations in one forum, where they can be consciously discussed and compared and where actors are forced – in the public light – to defend their views with arguments, is largely preferable in comparison to a more ‘automated’ process of mutually balancing forces. For it is only in such a forum that actors can achieve the kind of mutual recognition that is, according to James Tully, indispensable to constitutional dialogue in circumstances of profound difference and disagreement (Tully 1995). ³⁴

Conclusion: Political fiction?

The ideal-typical tripartite structure that we sketched above might read like a piece of political fiction. When practicing normative political theory, this is not necessarily a problem: the purpose of normative theory is not to describe reality but to explore models that real institutions and real actors should (but might not) try to realize in varying degrees of approximation. Nevertheless, we believe that the model we propose is not as detached from ongoing practices as it seems at first sight. In fact, several existing federal systems already embody fragments of a tripartite logic in their current procedures. Thus, rather than a mere utopian blueprint, our model can be seen as an attempt to capture this logic and offer a systematic reconstruction of it.

Where do we see this logic at work? For starters, the sheer fact that many federal systems have special procedures in place for constitutional amendments, procedures in which typically both the central government and the constituent units have a role to play, betrays a ‘tripartite’ concern. For what these procedures show, is that altering the federal architecture is something that should be neither in the hands of the central government nor in the hands of the constituent units. To put it in a different terminology, these procedures show that the Kompetenz-Kompetenz should not be located with the central government, nor with the constituent units. Are they not simply exercising the Kompetenz-Kompetenz in a joint manner? Yes, of course. But by doing so in a carefully crafted procedure that gives proper weight to both the central government and the constituent units, they are symbolically operating in a third place, that is, in a space that does not coincide with the regular political life of the central government, nor with that of the constituent units. They are constructing a place that does not belong to either of the two.

Really existent federal systems approach this tripartite logic in varying degrees. In federal systems where formal constitutional amendments are made within the central institutions, such as Germany, the space of constitutional politics is hardly distinguishable from that of regular politics. Even if the constituent units are represented at the central level, the procedures are clearly played out within the orbit of national politics. ³⁵ Similarly, when constitutional amendments are made simply among the governments of the constituent units, as is the case in the EU, there is clearly no attempt to create a space where the two levels of government are meeting as equals. However, when amending the federal constitution requires political action in the centre, but also in every single constituent unit (as in the United States), there clearly is a greater concern to set up a process that is beyond the scope of ‘normal’ politics. An even better approximation of tripartism, we argue, would be to have this process directed by an agency that is completely independent from either the central government or the constituent units. We would add that the closer a federal system approaches this tripartite ideal, the more truly ‘federal’ it is. When the process of adjusting the federal architecture is no longer located at the level of the central government, nor at the level of the constituent units, but is instead situated in a completely separate place, then the institutions finally match up with the federal idea that neither of the two levels can be considered a product or a dependent of the other.

The concrete example that approaches most clearly what we have in mind is probably the so-called ‘Canada Round’. Here, federal reform was truly given a space of its own, with a long timeline independent from regular electoral cycles and in places that were distinct from the regular centres of power. Moreover, there was a genuine attempt to include not only representatives of all constitutionally recognized entities but also to listen to the concerns of many other societal groups and to give a direct say to the broader public. The process was of course not perfect, yet we do find the example encouraging. In democratic times and when reordering the institutional architecture of our political world becomes a permanent activity, we see no real alternative but to look for ways in which processes of this type can be stabilized and be given a permanent home in solid institutions.

Again, very stable federal systems might see no immediate need for this. But we do think that our ideal model can provide inspiration for the institutional architecture of federal systems that are in need of reform. When the current procedural set-up does not provide suitable spaces to adjust the federal design, moving closer to a tripartite set-up might be an important part of the solution.