1. Introduction

Who has a claim to be included in a democratic polity? This has been a vexing question for political theorists as well as legislators and judges. Philosophers have tried to make the problem go away by adopting one of two contrasting strategies.

The first response is that democratic principles cannot resolve the problem and therefore we have to accept the historical contingency of political boundaries and the powers of nation-states to determine themselves who their citizens are. To be sure, most contemporary political theorists have added some critiques of current state practices or suggestions why some categories of individuals cannot be legitimately excluded from citizenship. Yet they often have done so starting from the premise that the context within which the question needs to be addressed is the international system of states as we know it.2 The problem is thus reduced to allocating territory and people to states in a way that does not challenge their boundaries and claims to self-determination.

The second response is to stick to a democratic principle and to use it for undermining the legitimacy of existing political boundaries. If boundaries are historically contingent, then they do not have deep moral significance and can also be radically questioned for the sake of democratic inclusion. Some theorists argue that the only democratically legitimate demos is a global one (Goodin 2007); others suggest that the demos ought to change depending on who will be affected by a particular decision (Shapiro 2000); still others regard democratic inclusion principles as norms that allow us to contest exclusion while not necessarily providing positive guidelines on how to construct alternative boundaries (Benhabib 2004, 2006 ; Näsström 2007).

The theoretical debate thus seems stuck between positions giving priority either to existing democratic boundaries or to principles of democratic inclusion that potentially challenge the legitimacy of all boundaries. But this standoff suggests already that there is something wrong in the way the debate has been framed. Since inclusion conceptually presupposes an external boundary, a theory of legitimate inclusion claims depends on a theory of legitimate boundaries. In other words, there is no point arguing for the right of individuals to be included in a particular demos if the legitimacy of that demos itself is either blindly accepted as a contingent result of historical processes or fundamentally rejected based on inclusion claims that are per se incompatible with drawing legitimate political boundaries.

The other reason for revisiting the democratic boundary problem after forty years of debate3 is that it simply does not go away in democratic politics even if philosophers try to conjure it away in democratic theory. Boundary and inclusion questions are among the most contested practical problems in contemporary democratic states. The rise of these problems on political agendas is arguably a result of democracies becoming more liberal and less self-confident in asserting quasi-natural boundaries of nation, territory and language. If the liberal transformation of democracy has contributed to making the boundary problem politically more salient, then the diagnosis that there is no cure for the problem that democratic theory can provide would be very bad news indeed.

Focusing on recent years in Europe alone, here is a small sample of events in which problems of democratic inclusion and boundaries have come up and had to be addressed by courts, legislators or by citizens in the election booth: the massive global trend of extending voting rights to citizens living abroad and a comparatively weaker European and Latin American pattern of letting non-citizen residents vote in local elections; an ongoing standoff between the European Court of Human Rights and the British government about the exclusion of criminal offenders from voting rights; the introduction of conditional ius soli in Germany in 2000 and Greece in 2010/20154 and the abandoning of unconditional ius soli by constitutional referendum in Ireland in 2004; the widespread introduction of language and civic knowledge tests as a naturalization requirement for immigrants in Europe since the late 1990s; the 2010 Rottmann decision of the Court of Justice of the European Union that member states have to take EU law into account when withdrawing nationality and the more recent moves in several EU states to deprive citizens joining a terrorist organization of their nationality; the Scottish referendum on independence in November 2014 and the nearly simultaneous rejection by the Spanish government and Constitutional Court of a similar referendum in Catalonia. All these decisions rely implicitly on contested ideas about democratic boundaries and membership claims. Normative theories of democracy need not be prescriptive in the sense of proposing specific answers for each of these issues, but they should at least be able to spell out the principles that ought to guide decisions. Yet many of the contributions to the democratic boundary debate seem keen to avoid this test.

This essay attempts to show that the diagnosis that there is no theoretical answer to the democratic boundary problem that would allow us to address its real-world manifestations is wrong. It takes the practical political manifestations of the boundary problem seriously by proposing that democratic inclusion principles must not only satisfy theoretical criteria, such as compatibility with broader principles of justice and democracy, internal coherence and answers to objections raised by rival theories, but also practical criteria that show how the proposed inclusion principles allow the boundary problems arising within democratic politics to be addressed.

My strategy is to argue that there is not a single principle of democratic inclusion but several principles, and that it is important to distinguish their different roles in relation to democratic boundaries. I also argue that polities into which individuals can claim to be included are of different kinds and it is equally important to distinguish the types of polity addressed by such claims. I do not argue, however, that there is an open-ended variety of inclusion principles or of kinds of polities and that inclusion always depends on context. That would be banal and undermine any effort at theorizing. The basic principles of democratic inclusion are limited and so are the basic types of democratic polities, and in my discussion I will reduce each of them to three. Such ideal-typical generalizations allow for identifying contexts where mixed principles apply or where polities are of mixed types.

The core normative argument of this essay is developed in section 3, where I discuss the principles of including all affected interests (AAI), all subject to coercion (ASC) and all citizenship stakeholders (ACS). I claim that these principles are not rivals but friends. They complement each other because they serve distinct purposes of democratic inclusion. Before this, I consider the general “circumstances of democracy” that consist in normative background assumptions and general empirical conditions under which democratic self-government is both necessary and possible. Section 4 contextualizes the principle of stakeholder inclusion, which provides the best answer to the question of democratic boundaries of membership, by applying it to polities of different types. I distinguish state, local and regional polities and argue that they differ in their membership character, which I identify as birthright-based, residential and derivative respectively. My conclusion is again that these are not alternative conceptions of political community but complementary ones. Each supports the realization of specific political values (of continuity, mobility and union) and taken together local, state and regional polities form nested democracies with multiple citizenships for all their members.

2. The circumstances and contexts of democracy

2.1 Diversity and boundaries

So how should we think about democratic boundaries? Neither as quasi-naturally given and beyond contestation, nor as features of a non-ideal world that we set aside when discussing what justice requires in an ideal world. Instead, we should think of boundaries as belonging to the circumstances of democracy. In his theory of justice, John Rawls defined the circumstances of justice as “the normal conditions under which human cooperation is both possible and necessary” (Rawls 1999: 109).5 We can describe political boundaries in the same way as belonging to the normal conditions under which democracy is both empirically possible and normatively necessary. Without claiming that these two conditions exhaust the circumstances of democracy, I suggest that democracy would not be necessary in the absence of a diversity of interests, identities and ideas, and would not be possible in the absence of boundaries.

In a society where all shared the same interests, a single collective identity as members and the same ideas about the common good, democracy would be pointless, since collectively binding decisions could be adopted unanimously or be taken by each individual on behalf of all others without any need for a procedure that aggregates their political preferences.6 Democracy is a system of political rule that provides legitimacy for collectively binding decisions and coercive government under conditions of deep and persistent diversity. Political ideologies that consider diversity as a non-ideal condition to be overcome through a transformation of society are therefore always potentially hostile towards democracy. This goes for orthodox Marxism and its ideal of a communist society without religion or economic competition as well as for nationalism and its ideal of matching the boundaries of cultural and political communities (Gellner 1983).

Boundaries are necessary background conditions for democracy for at least three reasons. First, without political and jurisdictional boundaries, democratic decisions would have indeterminate scope. This would be true even if every human being were included in a single global polity, since there would then still be a political boundary between human beings and other animals that could potentially be included.

Second, in the absence of political boundaries there is no distinction between intra- and inter-polity relations. This distinction is, however, constitutive for the political as a distinct sphere of human activity. Carl Schmitt's (1927/2007) friend–enemy dichotomy is just an extreme and implausible version of this distinction. Hannah Arendt expresses the democratic version of this argument:

Third, the existence of boundaries is a precondition for the democratic feedback mechanisms of voice and exit (Hirschman 1970). In the absence of any boundary, exit is by definition impossible. While easy exit may weaken the incentives for voice (in Hirschman's original “hydraulic model”), the absence of any possibility of exit fatally undermines the effectiveness of voice. A polity without boundaries is like a spontaneous crowd that has no addressee for voice, since it does not have collective procedures for counting votes and taking decisions.

These three arguments do not imply a defence of any existing boundaries. Instead, they suggest that we should imagine democratic citizenship always in a context where there is a plurality of other polities. The circumstances of diversity and boundaries can thus also be understood as referring to two sides of democratic pluralism: an irreducible internal plurality of interests, identities and political, moral and religious ideas, and an equally irreducible external plurality of political communities.

Although this is not essential for my argument, which focuses on democratic legitimacy, I believe that similar conclusions emerge for theories of justice. A vision of a world without political boundaries is dystopian in the same way as a world in which all human beings share a comprehensive moral perspective or the same way of life. The plurality of bounded political communities is constitutive for justice in the sense of forming a background condition against which questions about justice are raised.7 I am therefore inclined to think that political boundaries are also part of the circumstances of justice, which means that they have to be assumed as a background not only for non-ideal, but also for ideal theory. Political boundaries structure theories of justice fundamentally by subdividing them into three distinct sets of questions: justice within political communities (domestic), justice between political communities (inter-polity) and justice across political communities (trans-polity and global).8

Of course, theorists of global justice and cosmopolitan democracy generally do not imagine a single undifferentiated polity encompassing all human beings. What they intend to challenge is not so much the existence and utility of political boundaries but their moral status. They conceive of boundaries as instruments that allow for a top-down delegation of responsibility for specific territories and populations to particular governments (Goodin 1988, 2007 ) or the bottom-up aggregation of democratic votes in a global federation (Archibugi and Held 1995).

Philippe van Parijs summarizes succinctly the attitude of most global justice theorists towards political boundaries: “Nations, politically organized peoples … are sheer instruments to be created and dismantled, structured and absorbed, empowered and constrained, in the service of justice” (van Parijs 2011: 139). This view regards political boundaries and democracy itself as institutional arrangements whose legitimacy is entirely derived from how well they serve the goal of justice.

Our attitude will be different if we consider democracy as a set of institutions, the goal of which is to realize government of, for and by the people. In this view, popular self-government is a fundamental and intrinsic value, the pursuit of which must be constrained by requirements of justice, but which is at the same time a free-standing value that cannot be entirely derived from what justice requires. The primary purpose of democracy is to provide legitimacy to coercive political rule through popular self-government. While political boundaries should be regarded as a background condition for both justice and democracy, justice is not the same value as political legitimacy and may not always strictly require a democratic form of government.

Boundaries will then still be regarded instrumentally, but as a background condition that enables self-government. Particular boundaries remain open to contestation, for example if they are constructed in a way that denies some individuals full membership in a self-governing polity. Yet their democratic purpose is to create spaces of collective self-government of a people, which is incompatible with regarding the people itself as something “to be created or dismantled in the service of justice”.

My stance does not commit me to an essentialist conception of democratic peoples as nations. As I will argue in section 4, democratic peoples can be vertically nested within each other and also share horizontally overlapping memberships. At the same time, self-governing peoples must have the capacity to endow governments with comprehensive powers of agenda-setting and decision-making and to hold them also comprehensively accountable. Such peoples cannot be merely functional aggregates of individuals who happen to share an interest in a particular political decision or public good. Some theorists have suggested extending the idea of democratic self-government to “weak” or “functional” demoi whose scope is transnational and global, and varies with the decision at stake (Bohman 2007).9 While I will argue below that including externally affected interests is indeed a moral imperative for democracy, letting affected interests determine the boundaries of the demos would create indeterminate or ephemeral demoi that are structurally incapable of ruling themselves.10

To sum up my argument so far: The three reasons for assuming boundaries as background circumstances of democracy point towards a plurality of polities at all levels, including the global one. And asserting the intrinsic value of collective self-government points towards boundaries that demarcate comprehensive jurisdictions rather than issue-specific demoi. Taken together, these ideas are fully compatible with the project of cosmopolitan constitutionalism and the building of a global legal community (Habermas 2006), but exclude the vision of a self-governing global demos, even if we imagine it as federally or functionally subdivided into a plurality of dependent demoi.

2.2 Territorial jurisdiction and sedentary societies

Following Rawls's terminology, I have tried to identify transhistorical and transcultural circumstances that make democracy both possible and necessary. In a next step, I will now propose that a theory of democratic boundaries and inclusion must also take as given the fact that political boundaries demarcating comprehensive jurisdictions have territorial borders and that contemporary human societies tend to be relatively sedentary within these borders.

Unlike diversity and boundaries, territorial jurisdiction is neither a primary normative requirement for democracy nor a historically invariable condition. What we know about early human societies of nomadic hunters and gatherers suggests that their relation to territory was radically different from that of any political order after the Neolithic agrarian revolution. In our present world we do find non-territorial forms of democracy; some of them are institutionally established and complement a dominant territorial design of political rule,11 others flourish informally in the new virtual public spaces created by contemporary information and communication technologies. We can also imagine hypothetical future worlds in which territorial borders are much less relevant for democracy than today and individuals are identified as members of political communities based on non-territorial criteria.

Nevertheless, there are pragmatic as well as normative reasons for assuming that the dominant boundary structures of democracy are territorial. The pragmatic reason is that a theory of democratic boundaries would fail the “implications for democratic politics” test that I emphasized in the introduction if it remained at such a general level that it did not even take into account how democratic polities are territorially structured.

The normative reason is that territorial jurisdiction makes it more likely that democracy can emerge and be consolidated under the two circumstances of democracy. In relation to diversity, non-territorial boundary markers, such as shared descent, religion, political ideology, social class or ways of life, necessarily diminish internal diversity within such communities while enhancing differences between them. If comprehensively self-governing polities were primarily demarcated by these criteria rather than by territorial borders, democracy would be less needed since members would be preselected based on an assumed primary interest that they all share. At the same time, non-territorial polities would be so fundamentally dissimilar among each other that it would become very difficult to maintain support for any global legal order based on norms to which they all subscribe, let alone global solidarity and redistributive justice across such boundaries.12 The circumstances of both democracy and justice might thus be jeopardized in such a world. In relation to boundary stability, territorial jurisdiction has “lock-in effects” that make it more costly for political agents to exit and that strengthen therefore their motivation to exercise political voice (Rokkan and Urwin 1983; Bartolini 2005). If subjects can opt out of a political regime while retaining residence, they have strong incentives to free ride on public goods that can be accessed by all residents. At the same time, political entrepreneurs operating as rival authorities in the same territory will have incentives to rely on coercive extraction of contributions rather than democratic consent. By contrast, consolidated territorial jurisdictions, which need not be united under a single sovereign authority, create conditions under which subjects have reasons for preferring voice over exit and rulers have reasons to be responsive to their subjects.

Although territorial jurisdiction is a weaker and more variable condition for democracy than are diversity and boundaries, these arguments show that it is not a condition for non-ideal theory only. The tragic history of territorial conflicts between city republics, empires and nation-states should not delude us into assuming naively that territorial borders themselves are an obstacle rather than an enabling condition for democracy and peaceful relations between polities.

The territorial nature and borders of comprehensive jurisdictions provide a political-institutional background context for democracy. Yet there is also a closely related social condition that we need to spell out before we can address democratic inclusion problems. This is the assumption that territorial borders allow for categorizing human populations into residents and non-residents, with most of the laws that are adopted within a territorial jurisdiction applying to the former but not to the latter.

Human societies have different relations to territory that we can describe as static, nomadic, mobile or sedentary. In territorially static societies (nearly) all members spend (nearly) all of their lives in the territory where they have been born; nomadic societies, by contrast, move collectively through geographic space without ever settling down and taking up permanent residence anywhere; mobile societies share with nomadic ones the feature that (nearly) all members are constantly on the move but in mobile societies individuals move independently from each other, so that there is not even a collective relation to a territory based on the shared experience of joint movement.

The fourth type, which we can call sedentary, is a mixed one. It is a society most of whose members spend most of their lives in a particular territory (not necessarily the one where they were born). Sedentary societies are fundamentally structured around territorial residence. This does not imply that they cannot have members residing outside their territory or that everybody residing inside the territory is automatically a member. Instead, in sedentary societies territorial borders generate distinctions between immigrants, emigrants and natives that do not exist or cannot be distinguished at all in static, nomadic or mobile societies. Human societies since the invention of agriculture have been generally sedentary in this sense. My proposition is that we should accept relative sedentariness as a second background context for democracy.

Let me clarify this a bit further. First, the distinction between the four types of society depends on territorial scale. If we imagine the land mass of Planet Earth as a single territory, then all human societies have been static and will remain so as long as they do not colonize extra-terrestrial space or create swimming island polities in the high seas. If we shrink the territorial units of observation to sufficiently small size, then all societies have been mobile, since human beings are migratory animals who always move their locations of residence when observed over a sufficiently long time period. When examining instead a mid-range geographic and temporal scale, then patterns of human mobility have changed strongly over time, mostly from static and nomadic to sedentary and increasing levels of mobility since the onset of the industrial revolution. Our previous discussion of boundary structures as circumstances of democracy suggests that a normative theory should indeed assume mid-level territorial scales that encompass neither the whole globe nor are so small that comprehensive forms of territorial self-government would become impossible. Within this mid-range we will still find a wide plurality of types of territorial jurisdiction, from large empires and states to small municipalities.

This observation also makes it clear that in a nested multilevel structure of territorial polities, the degree of mobility that we observe depends on which level we use as reference and generally increases strongly as we move down from state to substate-regional to municipal level. This is so because internal borders at a higher level become external borders at the lower one. Municipalities have thus on average much higher percentages of immigrants and emigrants in relation to their sedentary populations than the provinces or states to which they belong because what counts for the state as internal migration is added to what the state classifies as international migration. We can thus describe multilevel polities without contradiction as simultaneously strongly sedentary and relatively mobile. In a multilevel polity, my normative proposition that sedentariness is a background context for democracy must therefore be specified as applying to the highest or strongest level of self-government. In a federal state, this level will be the federal one; in a union of states, it could well be the level of member states rather than that of the union. The condition of sedentariness should thus not be interpreted too strictly. I have suggested elsewhere that democracy would be difficult to sustain in hypermobile societies, which we can define as those in which at any point in time and at the strongest level of self-government a majority of citizens are non-residents and a majority of residents are non-citizens (Bauböck 2011b). This does not imply that a high volume of mobility that is contained as internal movement within a state or a union of states will undermine democracy in such a polity.

Second, I need to clarify why distinctions between immigrants, emigrants and sedentary populations, and the proportions between these, are normatively salient for democracy. One reason is the need to determine the personal scope of territorial jurisdiction. The concept of territory does not refer to land as a physical object, but to a geographically defined space within which political power is exercised over human beings and laws are applied to them (Buchanan 2004; Stilz 2011; Angeli 2015). In a hypermobile society territorial laws would apply mostly to transient populations whose primary political affiliation might be to some non-territorial political community. At the same time, if most citizens reside outside the polity and mobile residents escape legal duties by moving across borders, governments would have strong incentives to compensate for their loss of control over the population in their territory by expanding their extraterritorial jurisdiction and imposing ever more legal obligations (e.g. tax duties) on expatriates,13 which could eventually undermine the salience and stability of territorial political boundaries. A second reason is that democracy also needs a sense of “ownership” and belonging to the polity. It is difficult to imagine how hypermobile populations could be citizens of the territorial polity who authorize the government that issues and implements the laws to which they are subjected. If there is a relatively sedentary core population, then immigrants can integrate into the society while emigrants can remain connected to it across borders. Where there is no such core, it will be difficult to generate among territorial populations a sense of responsibility for the common good of the polity. Their moral obligations towards co-inhabitants will be the same as towards all other human beings outside the borders and the condition of subjection to a territorial government that they share with each other will be insufficient to generate perceived duties of solidarity, political participation or even just voluntary compliance with the laws.

Third, we should once again consider how a condition of relative sedentariness relates to the two circumstances of democracy, as we did with regard to territorial jurisdiction. On the one hand, both static and nomadic societies, by definition, lack substantive exit options and provide therefore inhospitable environments for democratic diversity and the interplay between exit and voice. Hypermobile societies, on the other hand, make it difficult to create comprehensively self-governing territorial jurisdictions, or make it more likely that such jurisdictions will be non-territorial, which diminishes internal diversity. My conclusion is therefore that relatively sedentary societies are a normatively salient condition for democracy, even though this condition has not always been present, may be lacking in particular contemporary societies and could be vanishing in a future hypermobile world.

A final observation is that the relation between the two territorial conditions for democracy is similar to that between the two circumstances of democracy. In both instances I have distinguished political-institutional features (boundaries and territorial jurisdiction) from social ones (diversity and relative sedentariness). And in both instances the two conditions are not independent of each other, but operate in tandem: boundaries distinguish internal diversity from an external plurality of polities; and territorial jurisdiction distinguishes internal mobility from external migration, while relative sedentariness creates the conditions under which citizens can collectively authorize and hold accountable a territorial government.

I have argued so far that we should address problems of democratic inclusion by assuming that democracies have boundaries of membership as well as territory. I have not argued that either of these boundaries must be sites where entry or exit is controlled and I have not argued that the two kinds of boundaries must match. In fact, I want to challenge both assumptions in section 4 of this essay: at local level, democratic polities have generally open territorial borders and purely residence-based inclusion, which implies that territorial and membership boundaries more or less coincide, while at state level migration leads to discrepancies between territorial borders and membership boundaries by generating non-resident citizens abroad and non-citizen residents domestically. The background assumptions I have made are weak. They provide a minimalist description of the world as it is and as we should assume it to remain even for the purposes of ideal theory. We need this background because principles of inclusion can only apply if and where there are boundaries that we consider at least potentially as legitimate. The distinction between the circumstances and the contexts of democracy serves a further purpose: while inclusion principles should only assume the circumstances of democracy, more specific norms of democratic membership depend on contexts that have changed over the course of human history and may still change in the future.

3. Purposes of democratic inclusion

After sketching the background for a theory of democratic inclusion let me now move to the foreground. Inclusion does not merely refer to the crossing of a boundary; it also implies a relation of correspondence between an individual or collective claim and an associative purpose. We normally do not say that criminal convicts are “included” in a prison, or that conquerors are “included” in the society they colonize because in these cases there is no correspondence between inclusion claims and purposes.

Let me explore this idea a bit further without developing a full conceptual analysis. The term “inclusion” strictly requires only one agent – the subject that includes – whereas the object that is included can be either an agent or a thing. A philosopher can include a particular argument in her analysis, a state can include a territory in its jurisdiction. However, democratic inclusion presupposes agency both on the side of those who are included and those who include them. This agency need not be expressed through explicit acts of consent. Families, states and most religious communities include those born into them without asking for their consent. In each of these cases inclusion is based on the notion of birthright. The term “right” makes it clear that there is a reference to a claim. The other agent in the relation is the association or collectivity that includes. A necessary condition for speaking about inclusion is that it serves a purpose pursued by this association. Where there are institutionalized rules for inclusion, they will reflect this purpose in the rules under which individuals or groups are included. A church will admit members based on their adherence to its doctrines and their contributions to the life of the congregation, a sports club based on their skills or their willingness to contribute to its budget and activities.

Neither need agency be expressed in explicit consent on the side of the including agent. We can illustrate the difference between consensual and automatic modes of inclusion by considering the contrast between birthright acquisition of citizenship, which is generally automatic, and naturalization, which requires an application. I will discuss this contrast and the specific purpose it serves in section 4. Here I am concerned with a general purpose of democratic polities, which is to achieve legitimacy for political rule. I propose that democratic inclusion principles specify a relation between an individual or group that has an inclusion claim and a political community that aims to achieve democratic legitimacy for its political decisions and institutions. Inclusion claims and purposes must correspond in such a way that satisfying the former is seen to contribute to the latter.

In the rest of this chapter I will elaborate a normative conception of democracy that relies on three distinct principles of inclusion, each of which serves a specific purpose of legitimation and operates within a specific perimeter. The principles are those of including all affected interests, of including all subject to the law and of including all who have a legitimate stake in membership. In contrast to most political theorists who have analysed the democratic boundary problem,14 I claim that the three principles differ in scope because they support different inclusion claims. Those whose interests are affected by a decision have a democratic claim that their interests be taken into account in the process of decision-making and implementation. Those who are subjected to the jurisdiction of a polity have a democratic claim to equal protection under the law. And those who have a legitimate stake in participating in the self-government of a particular polity have a democratic claim to be recognized as citizens.

It is tempting to imagine the territorial scopes of these three inclusion principles as concentric circles, with affected interests having the widest and citizenship claims the narrowest perimeter. This seems plausible when one considers that many political decisions have spillover effects across the borders of jurisdictions and that in immigration states there are often significant shares of non-citizens who are subjected to the laws in roughly the same way as citizens. However, this image of concentric circles is also misleading. International migrants are citizens of their states of origin and today they mostly retain rights to political participation and representation there while being only weakly affected by political decisions or subjected to the laws of these countries. Is there any justification for drawing the boundaries of membership wider or narrower than those of impact or subjection? This is a puzzle that I will address in section 4. Yet the question is less puzzling if we consider that inside the territorial jurisdiction of representative democracies, too, decisions are taken on behalf of all citizens but often affect the interests of a particular subgroup much more strongly than the rest of the citizenry. Consider, for example, a law that prescribes a certain curriculum in schools. If the law has merely regulatory and no fiscal impact, then elderly childless citizens can hardly claim that their interests are as strongly affected as those of citizens of minor age and their parents, yet the votes of the former and the latter will count equally in a referendum (or in a parliamentary election) that puts this issue on its agenda while children of minor age who are most directly affected will not be directly represented in this decision at all.

In past writings (Bauböck 2007, 2009 a and b, 2015b ) I have argued that the AAI and ASC principles are morally attractive but suffer from two flaws. They cannot resolve the democratic boundary problem because the boundaries they suggest are necessarily indeterminate and unstable,15 and they are polity-indifferent, which means that they generate the same prescriptions for inclusion in local, regional or state polities, although these polities require different membership norms. I have contrasted these and other democratic inclusion principles with an alternative principle of including all citizenship stakeholders that is sufficiently determinate in its practical implications and sufficiently flexible to support different inclusion rules for different types of polities. This argument was to a certain extent lopsided because it focused nearly exclusively on defending ACS as the appropriate principle for determining who should be recognized as citizens. I now want to explore in more depth the virtues of the AAI and ASC principles and why we should regard them as complementing my stakeholder account. But I will also try to show that AAI and ASC fail to meet the mark when they are considered as norms that by themselves cover the whole range of democratic inclusion claims.

3.1 Including affected interests

For each of the three principles we have to specify further their scope (inclusion of whom?) and their domain (inclusion in what?). With regard to scope, AAI can be interpreted in two contrasting ways: does it refer to actually affected or potentially affected interests? Robert Goodin has argued that the former interpretation is incoherent: “[W]hose interests are ‘affected’ by any actual decision depends upon what the decision actually turns out to be” (Goodin 2007: 52). In his view, interests are affected “not merely by the course of action actually decided upon”, but also by the range of alternative courses of action from which that course was chosen (ibid.: 54). He concludes that “[m]embership in the demos ought to extend to every interest that would probably be affected by any possible decision arising out of any possible agenda” (ibid.: 61–62).

This interpretation of AAI begs the question why an interest in agenda-setting should count as relevant for purposes of democratic inclusion. If we regard agenda-setting as the core power of a democratic legislator, then only those who have a right to authorize this legislator can be seen as having a legitimate interest in agenda-setting. Goodin's claim that only a global demos can be legitimate is thus derived from an implausibly wide conception of AAI that builds the conclusion already into the premise. A principle of including all potentially affected interests assumes from the very start the existence of a global demos whose members have a legitimate interest in participating in or being represented in setting a global political agenda. A democratically plausible interpretation of AAI must instead refer to interests in the choice between alternative decisions on an already set agenda (Owen 2012).

Goodin's theory may be implausible in its consequences but it is certainly coherent: the members of a self-governing demos (or the representatives they elect) must have agenda-setting powers rather than merely the power to pick a decision from an already set agenda. If AAI is the only valid principle for determining membership in a demos, then all those whose interests are affected by any possible decision arising out of any possible agenda must be included in the demos. A demos with agenda-setting powers formed under the AAI principle must therefore be global in scope. In order to refute the conclusion we must either adopt a weaker conception of democracy in which the demos does not have agenda-setting powers and is thus no longer self-governing, or we must reject the claim that AAI is the appropriate inclusion principle for determining membership in a demos. I propose to stick to strong democracy and drop the claim that AAI is the all-encompassing inclusion principle.

If we accept the circumstances of democracy and the plural structure of political boundaries, then the core power of agenda-setting (even for global political agendas) can only belong to particular demoi at the sub-global level. There will then be from the very start a distinction between those who have the power to set the political agenda and those whose interests are affected by political decisions. Persons whose interests are externally affected are those who suffer or benefit from a political decision without belonging to the group whose members have a legitimate interest in agenda-setting in a self-governing demos. Externally affected interests are, by definition, actually affected interests, since otherwise there would be no external boundary whatsoever that distinguishes a self-governing demos from those whom it might affect through its decisions.

The distinction between externally and internally affected interests is rather obvious once we consider that some of the most significant interests that democratic governments have to track emerge only because of the existence of political territories and government institutions. This basic fact explains also why government action normally affects the interests of those who reside permanently within its territory much more comprehensively than those living outside the border. We should thus not imagine political communities as being solely responsible for tracking interests that exist pre-politically, that is, before or independently of the structure of political boundaries and territorial jurisdiction.

While rejecting Goodin's claim that any attempt to limit democratic inclusion to actually affected interests is incoherent, David Owen (2012) has pointed out that a principle of actually affected interests must not be interpreted so narrowly that it refers only to the impact of the decision actually taken. Tracking affected interests requires taking these into account in decision-making, not after that decision has already been taken. Affected interests thus have a claim to be included in the process of deliberation that precedes the decision and not only the process of implementation that follows it. In other words, actually affected interests have a claim to voice. They must be heard and taken into account by those who take the decision. They form the relevant public for political decisions.16 Those whose interests are affected by democratic decisions, no matter whether they are citizens, subjects or completely outside the jurisdiction, have a right to justification of the decision that respects them as autonomous sources of valid claims (Owen 2011, 2012 ; Forst 2012).

It is obvious that the current international state system is deeply flawed in this respect. It is designed to reduce the duty of states to justify their decisions towards those on whom they impact outside their territorial borders and boundaries of citizenship. At most, it supports duties of justification towards other states. The representation of externally affected interests depends, then, on these being effectively represented by a government that has the power to confront the authorities of the state that takes a contested decision. Moreover, the mechanisms of intergovernmental representation of externally affected interests operate in most cases only ex post rather than in the run-up to the decision and thus do not satisfy the condition that actually affected interests must be heard before a decision is taken. If inclusion of affected interests is a requirement of democratic legitimacy, then the flaws of the international state system, which is not itself democratically structured, are no excuse for a democratic polity to ignore the interests of those who are outside its jurisdiction. Instead of delegating this task to a global demos, it must be regarded as one that each polity is morally obliged to address whenever the decisions on its agenda are likely to have significant external impact.17

The task could be met in different ways, of which intergovernmental consultations and negotiations is only one. A second, and increasingly important, response is government participation in the creation of regional or global governance institutions on issues that systematically spill across jurisdictional boundaries, such as climate change, refugee protection and the persecution of crimes against humanity.18 The third response is to directly represent externally affected interests in the decision-making process itself.

This might be done, for example, through transborder referendums on issues such as the opening or closure of nuclear power plants close to an international border. Note, however, that transborder referendums presuppose legitimately constituted separate demoi on both sides of the border. A simple majority in a referendum involving two polities in which each vote is counted equally on either side of the border is not a defensible decision rule since the outcome would be determined by the citizens of the larger polity. Creating instead two constituencies of equal size on either side of the border would involve highly arbitrary boundary decisions and would again fail to meet democratic standards of legitimacy since there would be no elected legislators who are accountable to voters for implementing their decision. A reasonable procedure requires thus separate majorities in both polities. The decision rule may vary depending on the nature of the issue at stake, but generally the most plausible one is that a majority in each polity must agree to any new policy that has significant impact on the other. Granting veto power over a political decision to the citizens of a neighbouring state is obviously a proposal for which it will be hard to get political support. My point here, however, is that doing so does not merge the two demoi into a single issue-specific demos, but retains their identities as separate and self-governing political communities.

The most ambitious idea is to include actually affected interests in the decision-making process through special delegates with voting power in legislative assemblies. Matthias Koenig-Archibugi has suggested a scheme of representing externally affected interests through delegates in each national parliament elected by all non-residents with seats allocated in proportion to the share of world income under the control of that state. In order to take account of the fact that all humans would be equally affected by the rules determining their external representation in every particular demos, he adds to this proposal a globally elected global constitutional assembly that determines these rules (Koenig-Archibugi 2012). By distinguishing coercively subjected individuals with full and equal citizenship claims from externally affected interests with partial citizenship (ibid.: 462), this model tries to combine the notion of a global demos constituted by affected interests with self-government claims of particular demoi. However, it is not obvious how externally affected interests could be involved in the self-government of every demos without undermining the very idea of self-government. Koenig-Archibugi's proposal aims at simultaneously achieving option-inclusiveness and agenda-inclusiveness. Yet if agenda-setting is a reserved power for the members of a self-governing demos, then letting the delegates of externally affected interests participate in determining agendas rather than in influencing decisions through deliberation generates over-inclusiveness with regard to membership.

Distinguishing between the two circles of inclusion could lead to alternative institutional proposals, such as issue-specific mandates for the representatives of externally affected interests to participate in consultative bodies. Even granting them votes in particular legislative decisions would still be compatible with the separation between agenda-setting and option-choosing powers. Extending a model developed for local jurisdictions within states (Frey and Eichberger 1999), such solutions have sometimes been advocated under the label of “functional demoi”, but this terminology is misleading if the issues and functions with regard to which a decision can be delegated to such a “demos” are themselves determined by a territorially bounded and agenda-setting demos.

My preliminary conclusion is thus that reasonable versions of AAI that respect the plurality of self-governing polities as a background condition cannot be accepted as comprehensive answers to the democratic boundary problem, since they fail to provide a principle for the legitimate constitution of such polities and claims to inclusion in them.

AAI theorists have replied to this objection that the principles of including all subject to coercion and of including all citizenship stakeholders can be easily restated as particular applications of AAI. In this view, AAI is the broader formula that encompasses the others.19 In Goodin's words, what really matters for constituting a demos is interlinked interests (Goodin 2007: 49). A reductionist strategy can thus simply include individuals’ interests in protection of their rights by a particular government as well as their interests in membership in a particular polity among those interests that governments have to track. While these are logically coherent moves that make ASC and ACS appear to be merely special versions of AAI, they obscure the essentially different normative claims that each of the three principles supports. Maintaining these differences is crucial for democracy and therefore we should consider the three principles as complementary to each other rather than as broader or narrower versions of the same principle. This requires clarifying that, as a democratic inclusion principle, AAI refers specifically to interests in policy decisions rather than to interests in rights protection by government institutions or to membership in a political community. These last two interests are the domains of the ASC and ACS principles respectively.

3.2 Including the subjects of coercion

Democratic governments have special responsibility for those whom they govern. They must treat them with equal respect and concern (Dworkin 1977) and must secure their basic rights and freedoms. This is at least true for all liberal versions of democracy under the rule of law. Government is by its very nature coercive. The ASC principle captures the idea that the democratic legitimacy of government coercion depends on securing equal liberties for all whose autonomy it restricts.

This principle differs from AAI in important ways. First, it distinguishes between those who are subject to government coercion and those who are not, and it attributes special inclusion claims to the former only. Not everyone whose interests are affected is subjected to coercion. Governments can legitimately enforce laws only within a territorially limited jurisdiction and to some extent also over their citizens outside the territory. Recent developments in human rights have created universal jurisdiction of national courts for a strictly limited set of issues, such as crimes against humanity. These are exceptions that confirm the rule. At least prima facie, the scope of ASC is limited by the scope of jurisdiction. Second, ASC entails a stronger claim to equality. As explained above, the impact of democratic decisions on individuals’ interests is notoriously unequal and varies from one decision to the next. By contrast, ASC refers to subjection to government institutions rather than exposure to particular legislative outputs. Residents within a territorial jurisdiction are unequally affected by government decisions but – with minor exceptions, such as diplomats or tourists – equally coerced by government institutions. This suggests, third, that the proper domain of inclusion under ASC is representation of their interests in protection of their rights within government institutions.

The first of these features, the limited scope of ASC, means that the principle is better adapted than AAI to what I have called the circumstances and territorial contexts of democracy. However, the problem with ASC in this respect is that it is systematically biased towards existing boundaries. If inclusion claims are derived from the scope of current jurisdiction, then ASC may be too conservative in taking for granted borders as they are.

Some decisions taken by governments not only affect external populations’ interests but also subject them comprehensively to coercion in a way that fundamentally restricts their autonomy. This is most obviously true for military interventions and explains why these are legitimate only for purposes of self-defence or for humanitarian reasons. In such cases, the ASC principle would support our moral intuitions that states engaging in such forms of extra-jurisdictional coercion have special duties to admit refugees. It would also require that military authorities exercising coercive rule in a foreign territory have to treat the civilian population there with equal respect and concern. But how can ASC account for our intuition that it was legitimate for the U.S. to occupy German territory after World War II, while it would have been illegitimate to annex it (Stilz 2011: 590–591)? The answer must be that the overriding duty of democratic states engaging in military interventions is not one of inclusion; it is a duty of non-interference with the local population's right to self-government.20 Limiting the protective duties of democratic governments to populations within their jurisdiction is thus necessary to avoid the citizens of independent polities being curtailed in their rights of self-government.

Consider now how this would apply to colonial contexts. Until independence in 1962, Algeria was legally incorporated into the French territory. If ASC is interpreted as a claim for membership inclusion, then the colonial subjects in Algeria had a right to equal citizenship in France rather than to independence from France. A democratic principle of membership must link individual inclusion claims to collective self-government claims in order to avoid a status quo bias in favour of unjust territorial borders and jurisdictional boundaries.

I share therefore Anna Stilz's conclusion that territorial claims of states must refer to a conceptually prior relation of a people to the territory. Yet Stilz falls back on a statist version of the ASC principle when she defines peoplehood in terms of “a history of political cooperation together by sharing a state (legitimate or otherwise) in the recent past” and the political capacity of the group “to reconstitute and sustain a legitimate state on their territory today” (ibid.: 591). The implication seems to be that colonies without prior statehood would fail the political history test and Algeria would thus have had no claim to independence rather than inclusion in the French polity. Stilz tries to avoid this conclusion by acknowledging that “[e]ven if a group has only shared an illegitimate state in the past, they may qualify as a people if that shared history has been combined with other joint activities that have created the political capacity for them to sustain a legitimate state today” (ibid.: 592). Yet this somewhat ad hoc adjustment remains stuck in the logic of ASC by making self-government claims entirely dependent on prior subjection by a state: “only a history of sharing a state demonstrates the existence of the moral bonds that support political authority” (ibid.: 593). This logic is pernicious for indigenous peoples, who would generally fail both the historical and the capacity tests since their territorial self-determination claims rely neither on a history of, nor a capacity for, independent statehood. Although Stilz is on the right track when insisting that we need an account of peoplehood in order to justify claims to territorial jurisdiction, her view remains too narrowly statist both retrospectively (the political history test) and prospectively (the political capacity test). What we need instead in order to avoid the potentially oppressive over-inclusiveness of ASC is a prior conception of political membership linking individual inclusion claims to collective claims to self-government and a conception of political community that is not limited to sovereign states.

Let us now consider the scope of inclusion under ASC in contexts where current jurisdictional boundaries can be regarded as legitimate. Equal protection must then be offered to all residents but territorial scope ought to be interpreted with some flexibility. The duty of equal protection for all within the jurisdiction needs to track the impact of being subject to coercive legislation on individuals’ freedom. While tourists will hardly qualify, temporary migrants may experience significant restrictions of their autonomy, especially if they do not enjoy the same freedom of movement and legal protections as long-term residents. Citizens and residents who are temporarily outside the jurisdiction are clearly covered by ASC and so must be coerced emigrants who have been driven into exile by a non-democratic predecessor regime. In these cases, the situation of individuals is comprehensively marked by subjection to coercion that they have experienced in the past and this creates an ongoing duty of protecting their rights.21 The situation of emigrant citizens who live permanently abroad is less clear in this regard. In the standard case they will have interests affected by legislation concerning their diplomatic and consular protection, their right to return, and their property rights and tax duties in countries of origin. Yet they are not comprehensively subjected to the legal order of these countries. Claudio López-Guerra insists therefore that voluntary long-term emigrants should lose their citizenship status and voting rights (López-Guerra 2005) while David Owen argues that they remain subjected to those laws that concern their status as citizens abroad as well as to the general constitutional order, and have thus at least a claim to political participation in constitutional referendums (Owen 2010). My own view is that ASC does not provide a solid basis for citizenship status rather than some form of partial membership for long-term emigrants and their offspring born abroad. Their claims can be better understood from a stakeholder perspective applied to the specific context of citizenship in the international state system.

Most ASC theorists invoke the principle in order to determine who has a claim not only to equal protection but also to citizenship status and voting rights. These are two different questions, the answers to which need not be derived from the same inclusion principle. The distinction is clearly drawn in section 1 of the Fourteenth Amendment to the U.S. Constitution:

The “citizenship clause” of the Fourteenth Amendment specifies a set of rules for determining who are the citizens of the polity. These are territorial birthright, naturalization, and linkage between state and federal citizenship. It also includes a “subject to the jurisdiction” condition that has in the past been invoked restrictively in order to permit citizenship revocation for naturalized U.S. citizens who took up permanent residence abroad. After World War II, the U.S. Supreme Court reinterpreted the Fourteenth Amendment in such a way that today U.S. citizenship can only be lost through relinquishment or voluntary renunciation (Aleinikoff 1986; Weil 2013). While the “privileges and immunities” clause still refers specifically to citizens, the two subsequent clauses speak about persons rather than citizens. The distinction is intended and important: due process of law and equal protection of the laws are owed to any person within U.S. jurisdiction, not only to those whom the first clause identifies as U.S. citizens. Conversely, not all who have a claim to due process and equal protection have thereby also a valid claim to citizenship.

This is not yet a conclusive normative argument. First, there is no reason for privileging the U.S. Constitution as an authoritative source for interpreting democratic inclusion over other constitutions or international legal documents. Second, the rules listed in the citizenship clause are legal mechanisms rather than normative principles. So we still need to find normative reasons why those who have a claim to equal protection should not also have a claim to citizenship status. This will be the task of the following section on stakeholder citizenship.

Before embarking on this task, we need to consider further the domain of inclusion under ASC and which institutional arrangements can track the interests of those subject to coercion. The potential problem with the equal protection claim is that it does not explain why democratic governments can be trusted to secure the freedom of those whom they coerce. The democratic answer to the old question: “who will watch the watchmen?” must be: “those who are watched by the watchmen”. In order to qualify as democratic, the protection of rights through government institutions must include rights to contest government authority and decisions. In other words, democratic institutions must not only provide equal protection to those subjected to them, but must also open up channels for contestation.

The contemporary version of republicanism that has been called “neo-Roman” and is best represented by intellectual historian Quentin Skinner and political theorist Phillip Pettit provides a solid normative grounding for this interpretation of the ASC principle. Pettit has put special emphasis on contestability as a condition for non-domination. He distinguishes between “authorial” and “editorial” control of governments and tends to privilege the latter: “Government will be authorially controlled by the collective people under electoral arrangements whereby issues are decided by plebiscite or representatives are chosen to decide them. And government will be editorially controlled by the people under arrangements of a broadly contestatory kind” (Pettit 2006: 3). Both types of control are relevant for democratic legitimacy but the collective people that exercises them is not necessarily the same. The normative claim to be able to contest government institutions and their decisions follows directly from being coercively subjected to them, whereas the claim to be included in the demos that exercises “authorial control” does not. The very term “control” is also somewhat misleading in the latter context, since governments have to be authorized through popular vote before they can exercise power that is then controlled by the people.

Pettit's theory is in this respect at odds with other republican traditions, such as Rousseau's, Kant's, Arendt's and Habermas's, that have emphasized the link between individual and collective self-government. Pettit's exclusive focus on domination defined as vulnerability to arbitrary interference that fails to track one's interests (Pettit 1997, 2012 ) risks losing sight of the regulatory ideal of popular sovereignty and its – always imperfect – realization through democratic procedures for electing – rather than only controlling – governments. This shortcoming makes neo-Roman republicanism a somewhat limited perspective for a comprehensive theory of democratic inclusion, but one that has elective affinity with the ASC principle and is well suited to provide normative support for it. Although Pettit's general description of domination seems to be compatible with AAI, vulnerability to arbitrary interference is a condition in which individuals find themselves as the result of exposure to coercive government institutions rather than to negative externalities of particular decisions.

In contrast with the representation of externally affected interests in political deliberations and decisions, the institutional devices for securing equal protection of the law and opportunities for contestation are conventional and do not have to be newly invented. They include constitutional protection of fundamental rights and judicial review of ordinary legislation by constitutional courts as well as institutionalized complaints and contestation procedures for individuals in courts and ombudsman bodies and, finally, the rights to protest against governments and their decisions through political speech and activities.

From an inclusion perspective the important question is who should be protected and have access to contestation opportunities. If the answer is: all subjected to government jurisdiction, then citizens and non-citizen residents must enjoy these rights equally. This is not only implied by the U.S. Fourteenth Amendment. In Europe a principle of non-discrimination on grounds of nationality with regard to protection of fundamental rights is enshrined both in the European Human Rights Convention and in EU law.22 Yet the scope of such protection is continuously under dispute. Concerning contestation rights, restrictions on aliens’ rights of political association and activity were very common throughout the twentieth century and linger on in a number of contemporary European states with regard to rights to membership in political parties.23

On some interpretations, however, the scope of ASC is even wider than this. Arash Abizadeh has argued that immigration control is coercive towards non-citizen non-residents in a way that gives them a claim to be included in the demos for the purpose of making legitimate immigration law (Abizadeh 2008, 2010 ). The idea that outsiders have a right to participate in the making of the very laws that exclude them is a good example of how standard versions of the democratic boundary problem can generate rather perplexing paradoxes.24 David Miller's reply to Abizadeh is that would-be immigrants are not coerced if immigration control merely removes one option from their choice set of potential destinations (Miller 2010). If one accepts Miller's response, one may still hold on to ASC while avoiding the consequence that immigration control subjects the rest of the world to coercion and generates thus a global issue-specific demos for purposes of regulating immigration.

Yet even under Miller's interpretation, very significant numbers of actual rather than potential migrants could claim to be coerced because they do not have sufficiently robust opportunities to choose alternative destinations. This argument would apply to refugees as well as to dependent family members of immigrants and possibly also to large numbers of poverty- or environmentally driven migrants. Abizadeh's argument has the merit of drawing our attention to the fact that coercive jurisdiction is not only exercised within territorial borders, but also that borders themselves are potentially coercive instruments if they are not only used for demarcating jurisdictions but also for controlling migration flows. No matter whether this happens at the border or coast, inside the territory, at the high seas or through “remote control”, points of departure in other countries, it is hard to see how state responsibility for the protection of migrants whom they turn away – and who have stronger claims to admission to this country than anywhere else – could be denied under any plausible version of ASC.25 This suggests a somewhat more expansive scope of the ASC principle but does not yet support Abizadeh's conclusion that immigrants have to be included in the demos already before entry. This latter conclusion relies on interpreting ASC as a membership principle, which is what I intend to question.

I close this section by pointing to another problem with such an interpretation of ASC: it is likely to support not merely voluntary but also mandatory inclusion in the demos. Since the end of the nineteenth century, states have generally refrained from naturalizing first generation immigrants against their will. Yet if a legitimate demos has to include all who are subject to political coercion, then foreign nationals belong to it already by virtue of residing within the jurisdiction. So it is not clear why they should have a freedom to opt out while staying in the territory. At the same time, democracies may have relevant interests in naturalizing foreign residents. They may lack legitimacy if they rule over large numbers of foreign residents who cannot participate in elections. A persistent internal boundary between foreigners and citizens may also undermine social cohesion and a sense of joint responsibility for the common good. Finally, foreigners who refuse to naturalize may do so in order to free ride on public goods to which only citizens contribute (such as military defence where citizens are drafted). All these are serious concerns and the idea of mandatory citizenship for immigrants has therefore found a few defenders among political theorists (Rubio-Marín 2000; Carens 2005; López-Guerra 2005; de Schutter and Ypi 2015).

As I will argue in section 4 of this essay, the problem is not with the idea as such, but with its application to the specific context of citizenship in the international state system under conditions of relative sedentariness. It is in this context that mandatory citizenship for immigrants clashes with their right to choose between alternative citizenship statuses as well as with mutual obligations between independent states. The problem with ASC is not that it supports automatic acquisition of citizenship based on residence within a jurisdiction, which is entirely appropriate for citizenship at the local level, but that it does not allow for distinguishing between national and local contexts.

3.3 Including citizenship stakeholders

The versions of AAI and ASC that I have outlined and supported above cannot be accepted as sufficient for democratic inclusion because they could potentially be fulfilled also by non-democratic regimes. Imagine an enlightened and benevolent autocratic government that does not have any democratic mandate to rule but whose sole aim is to govern well according to liberal standards of justice. Such a government would fully take into account all interests that are actually affected by policies that it has put on its agenda and would invite representatives of groups whose interests could be impacted to its deliberations. Its decision would then be based on a careful calculation in which these voices are weighted by the strength of preferences and potential impact. The government would also make sure that it protects the full panoply of civil and social rights for anybody in its jurisdiction while refraining from interfering in this regard with the jurisdiction of neighbouring governments. It would expose its policies to strong forms of judicial review by independent courts, would provide multiple opportunities for complaints and would not curtail freedom of political speech, association and activity. But this liberal government would be appointed by an enlightened monarch. In the absence of a popular mandate we would not call such a government democratic.

Proponents of AAI and ASC will of course protest that their versions of the two inclusion principles are democratic because – in contrast to my account – they are not limited to giving voice to affected interests and protection to the subjects of coercion, but identify also the members of the demos who elect the government and to whom it must be accountable. Yet if we stick to a strong version of democracy as popular self-government, then we end up in circular reasoning: the demos is constituted through the impact of a government that can itself only be constituted through the very same demos.26 In order to avoid this circle, we would have to accept that governments create democratic peoples rather than the other way round.27 Elections would then be just another device for “editorial control” over governments rather than an original source of their legitimacy. And when constitutions invoke the “people” as the ultimate constitutional law-giver this would be little more than empty political rhetoric. So it seems AAI and ASC can maintain their democratic credentials only at the price of accepting weak versions of democracy that abandon the normative ideal of self-government and replace it with a supposedly more realist conception according to which democracy is about controlling government power rather than authorizing it. Since such control comes ex post, the role of the demos is then effectively similar to that of an independent judiciary.

If we understand citizenship instead not merely as a bundle of rights and duties but as a status of membership in a self-governing polity, then a democratic inclusion principle that determines who has a claim to citizenship must focus on the relations of individuals to a particular political community rather than to a government and its decisions. The stakeholder principle that I have defended is only one among several inclusion principles that satisfy this formal condition. Elsewhere I have distinguished pre-political and political principles (Bauböck 2015b). Democratic and liberal nationalists propose that the political community should be understood as a nation with a historically stable cultural and territorial identity. The nation is a pre-political community not because of its origins, since national consciousness is often the result of nation-building policies pursued by states. It is pre-political because of the cultural content of shared identity and the corresponding criteria for membership. Constitutional patriotism is not enough (Kymlicka 1995, 2001b ). In order to be included in the imagined community of the nation, individuals must either be born into it or adopt cultural repertoires shared by its members. By contrast, social contract theories that imagine the political community as a voluntary association of individuals are political in the sense that they ground the identity and legitimacy of a polity in a shared political purpose. The problem with both ideas is that they are difficult to square with liberal conceptions of democracy.

From a nationalist perspective, admitting new members to the political community must serve the purpose of nation-building. In some historical contexts, this meant inviting immigrants from diverse origins; in others it meant selecting them on the basis of presumptive cultural fit or prior national membership; in still others it meant closing the borders and denying access to citizenship for the sake of preserving national identity. If political community can only be achieved through nation-building, then the inclusiveness of the polity for minorities and newcomers is not a matter of democratic principle but of historically contingent circumstances. A nationality principle may be compatible with liberal democratic standards in some contexts but will conflict with them in others. It can thus not serve as a general guideline for democratic inclusion.

The idea that a democratic polity should be imagined as a voluntary association of citizens leads to similar problems. A voluntary association is self-governing if its members are not only free to exit, but also to admit or exclude outsiders who are willing to join. This is a political and at first glance also democratic conception if it leaves decisions about membership to democratic procedures that track the collective will of current members. As pointed out by Robert Dahl when rejecting Schumpeter's claim that “we must leave it to every populus to define himself” (Schumpeter 1942/1976: 245), this would entail that any polity is democratic as long as it is governed by a body that is internally democratic even if it excludes the vast majority of those subject to its laws (Dahl 1989: 121–122). Christopher Wellman's conclusion that states deriving their legitimacy from a principle of voluntary association and dissociation are free to select immigrants on grounds of race and “entitled to reject all potential immigrants, even those desperately seeking asylum from corrupt governments” (Wellman 2008: 141) must be equally disturbing for liberal democrats.

The stakeholder principle starts from a different conception of political community that consists of both empirical and normative assumptions. I have already spelled out the empirical assumption when introducing the circumstances of democracy: a plurality of bounded political communities is part of the human condition. Humans are social animals. They have strong stakes in being recognized as members of particular political communities because being an outsider who does not belong to any such community is a condition of extreme precariousness. To put it positively: membership in a polity is a necessary condition for human autonomy and well-being. It is not, however, a sufficient condition because political rule can also destroy freedom and deprive people of their subsistence, as it has done in most of its manifestations over the course of human history. A person's stake in being a member of a particular polity depends thus on that polity being governed in a way that protects its members’ autonomy and well-being. So far, this seems like another version of the ASC principle. The democratic twist comes with the further assumption that those who have an interest in protection of their individual freedom and well-being by a particular polity thereby share with each other an interest in the collective freedom and flourishing of that polity. Citizens are stakeholders in a democratic political community insofar as their autonomy and well-being depend not only on being recognized as a member in a particular polity, but also on that polity being governed democratically. Political legitimacy in a democratic polity is not derived from nationhood or voluntary association but from popular self-government, that is, citizens’ participation and representation in democratic institutions that track their collective will and common good.

Unlike AAI and ASC, ACS derives inclusion claims from a correspondence between individuals’ interests in autonomy and well-being and the collective interests of all citizens in their polity's self-government and flourishing. The term “stakeholder” can be easily misunderstood as referring to a stake in particular democratic decisions or in the protection of one's rights by a particular government (see for example Beckman 2009: 41). It should instead be understood as having a stake in membership, which is why I refer to “citizenship stakeholders”.

Assuming that individuals have a general interest in membership in a self-governing polity is much less demanding than those versions of republicanism that attribute intrinsic value to political participation. As pointed out by Will Kymlicka, this ideal is unattainable in contemporary liberal states. It conflicts with liberal toleration of a plurality of conceptions of the good and can serve to justify citizenship exclusion based on lack of civic virtue (Kymlicka 2001a: ch. 7). The idea that membership in a self-governing political community is a universal and intrinsic value is, instead, fundamentally inclusive and compatible also with ways of life that are inherently apolitical. Even the members of reclusive monastic orders will be better off as citizens of a democratic polity than as stateless persons or as subjects of autocratic rule.

At the same time, and unlike AAI, ACS provides normative reasons against over-inclusiveness. The relation between individual and collective self-government is bidirectional. Individuals have a claim to inclusion if their autonomy depends on the collective freedom of the polity. But the polity can also reject the inclusion of non-stakeholders on grounds that it would undermine the capacity of citizens to govern themselves. One illustration for over-inclusiveness is provided by current laws in many European states that allow for unlimited descent-based transmission of citizenship status across generations born abroad. Such perpetual ius sanguinis generates selective openness based on national origin for immigrants who have no stronger stake in the polity than others that are denied entry. Moreover, since in a majority of democratic states citizenship status is today sufficient for exercising the franchise, under such a rule the votes of outsiders are counted together with those of citizenship stakeholders, which infringes the self-governing rights of the latter.28

The reference to collective claims to self-government means that, unlike ASC, ACS is not inherently conservative with regard to existing borders. If these borders prevent a particular political community from governing itself, they infringe thereby also on the claims of individuals to citizenship in that community. The question of how conflicting claims to self-government can be sorted out is too complex to be fully addressed here. Accepting a plurality of polities as part of the circumstances of democracy entails that collective self-government claims are legitimate only insofar as they accept territorial and membership arrangements that allow for simultaneous fulfilment of all legitimate claims by other polities. Such a “compossibility principle” will in many cases lead to territorially nested forms of self-government, in which the legitimacy of national majority claims to self-government within a territory that includes national and indigenous minorities will depend on accepting a constitutional identity of the state as a plurinational polity and territorial autonomy for these minorities (Kymlicka 2001b: ch. 5; Tully 2001, 2008 : vol. I, ch. 6). Conversely, under these conditions territorial minorities can be morally and legally obliged to respect the territorial integrity of a state that grants them sufficient autonomy and accepts them as constitutive communities in a plurinational polity (Bauböck 2000, 2002 ). Nested self-government is not always possible, however. Where territorial populations are oppressed as external colonies or where internal minorities are excluded from majority nation-building projects, they have remedial rights to self-determination that may result in the formation of autonomous territories or new independent states.29

Compossibility is a side-constraint on legitimate self-government claims, but it is not sufficient to sort out spurious from genuine claims. The condition in this regard is that genuine claims must be representative of either an existing citizenry in a legitimately established polity or of a potential citizenry that has manifested a desire for collective self-government for some time and has experienced the absence of self-government institutions as a form of oppression. In the former case, self-government claims may be directed against attempts to curb or dismantle current powers of territorial autonomy; in the latter they may support devolution or – in the extreme case – independence. Testing the representativeness of claims is potentially as difficult as assessing compossibility.30 My point here is, however, a theoretical one. A representativeness condition provides us with a political and democratic account of the legitimacy of peoplehood as an alternative to nationalist and associative explanations. From this perspective, a democratic people is created through representation of its claims to self-government. Democratic representation is not merely a procedural device through which citizens can control a government that rules over them; the democratic people itself is constituted through representation of its claims to self-government.

The legitimacy of jurisdictional boundaries is a question that must be solved before individual inclusion claims can be addressed. This makes the question of who ought to be enfranchised in plebiscites about territorial independence or border changes, such as the 2014 Scottish referendum, a particularly perplexing one (Ziegler, Shaw and Bauböck 2014). Once territorial borders can be regarded as democratically legitimate because they are uncontested, because they enable simultaneous nested self-government or because they are the just outcome of remedial exercises of self-determination rights, then the claims of individuals to inclusion in a particular polity can be decided on the basis of a “genuine link” principle.

“Genuine link” is a doctrine in public and private international law that is invoked to establish or dispute the right of states to award their nationality and to grant diplomatic protection to or impose duties on individuals whom other states also claim as their nationals.31 For purposes of democratic theory, genuine link can serve instead as a critical standard for assessing the strength of ties between an individual and a particular polity. Keeping in mind that the substantive purpose for citizenship attribution is to secure, on the one hand, political conditions for individual autonomy and well-being and, on the other hand, collective self-rule, this strength cannot be measured in a uniform way either as a subjective sense of belonging or through objective indicators such as duration of residence or family ties in the territory. What counts as genuine link depends also on the nature of the polity itself and will be different for citizenship in states, municipalities and regions. Having been born and raised in the jurisdiction will generally be irrelevant for claims to local citizenship, whereas it may be sufficient for claims of first generation emigrants to retain their nationality of origin and the right to return.

Let me conclude this section by considering a major challenge for a stakeholder perspective. Even if the link between individual autonomy and collective self-government need not imply that citizens have a duty to participate actively in the political life of the polity, it does imply that they must have the opportunity to do so. But this opportunity in turn depends on their capacity to participate. The citizenship status of minor children or cognitively disabled persons might then be in jeopardy under this conception, whereas AAI and ASC would have no difficulty in arguing for their inclusion. For AAI, individuals must be capable of having interests, which presupposes sentience, a sense of selfhood and capacity for purposive action. As Donaldson and Kymlicka point out, not only minor children and most cognitively disabled persons but also many animal species meet these criteria (Donaldson and Kymlicka 2011). Policies could then be regarded as democratically inclusive for these human and non-human individuals if they track their interests through mechanisms of indirect representation. ASC may be somewhat more demanding if we distinguish coercion from harm by presupposing an individual will (in the sense of a capacity for forming and pursuing projects) that can be coerced. Still, even this condition does not plausibly draw a line between intelligent human and non-human beings, nor between minor children and mentally disabled humans, on the one side, and adults, on the other side. Does a view of citizens as stakeholders in a self-governing polity exclude any or all of these categories?

The case of minor children seems easiest, since they are expected to develop the cognitive capacities that will allow them to participate politically. Children's rights activists have, however, rejected a view according to which children are merely in need of protection and without capacity to participate. Instead, they see them as having a right to participate according to their capacities in shaping the conditions under which they live and grow up (Donaldson and Kymlicka 2015). This cannot mean that minor children have a claim to voting or standing as candidates in legislative elections. And giving parents proxy votes that they can cast on behalf of their minor children looks more like a violation of the one-person-one-vote principle in favour of a particular category of adults than a vehicle for children's participation in the polity. If there is no democratic way of providing children below a certain age with opportunities for participating in electoral politics, does this mean that from a stakeholder perspective they are just partial citizens? And what about newborn children? How can they be considered citizens rather than merely as persons whose interests and well-being democratic states have a duty to protect?

I think the answer lies in the conditions for continuity of a self-governing polity over time. Newborn babies are attributed citizenship not just because we regard them as future citizens. If this were the case, one might as well wait until they have reached the age of majority and consider them until then subjects within the jurisdiction who have a claim to equal protection.32 The reason why we recognize them as citizens is that political communities are transgenerational human societies. The status of membership in such communities is acquired at birth and does not depend on age-related cognitive or other capacities. In democracies, it is the larger transgenerational society that collectively governs itself and not the subcategory of adults who have the capacity and opportunity to vote or hold public office. Minor children are citizenship stakeholders because of their belonging to a transgenerational political community.

This does not mean that voting rights are irrelevant to democratic inclusion. They are the most important power that citizens hold equally and restrictions of access to this power must be justified. The benign liberal autocracy that I introduced at the beginning of this section treats adult citizens as if they were minor children and this is deeply degrading no matter how wise and benevolent the decisions taken by the government. By contrast, there is nothing degrading about treating children as children, which includes responsibilities to allow them to participate in all decisions concerning them. For adults this is not enough. In democracies, governments do not allow citizens to participate politically insofar as they are fit to do so. It is the other way round: citizens allow governments to govern them insofar as governments are fit to do so.

A stakeholder conception does therefore suggest a distinction between the demos, consisting of all those who have the franchise, and the citizenry, composed of all who have a stake in being members of a transgenerational political community. This distinction does not exclude either minor children or mentally handicapped persons from citizenship on grounds of cognitive incapacity. What it does, however, is cast some doubt on the idea promoted by Donaldson and Kymlicka that domesticated animals should also be considered citizens (Donaldson and Kymlicka 2011: ch. 5). It is hard to reject this idea on the basis of AAI and ASC and it is difficult to accept it from a stakeholder perspective. As Donaldson and Kymlicka explain convincingly, domesticated animals can and should be treated with respect and concern as members of the oikos. Duties of care towards such animals should also be backed by public policies and laws. But the plurality of bounded and transgenerational political communities is part of the human and not of the animal condition. Membership in such communities is therefore species-specific. Challenging this political boundary will do little to improve the conditions of domesticated or other animals and might do great harm to the idea of equality of membership that is fundamental for democracy.

3.4 Synthesis and tensions

I have examined three different types of democratic boundaries: those marking the impact of political decisions, boundaries of government jurisdiction and boundaries of membership in a self-governing polity. For each of these there is a corresponding principle that identifies individuals who have claims of inclusion: those whose interests are affected, those who are subject to government coercion and those who have a stake in citizenship. Finally, each of these principles includes individuals in different democratic activities: in the deliberation and decision of policies, in the protection by and the contestation of political authority, and in the authorization of governments and of specific policy decisions through democratic elections and referendums. Since democratic inclusion serves different normative purposes, there is no compelling reason why the three boundaries ought to be congruent. Their incongruence ought to be accepted as a permanent feature of democracy in a world with strong interdependence and migration flows between autonomous polities. Trying to bring about congruence would be harmful for democracy. Expanding the boundaries of jurisdiction and membership to the widest circle of all affected interests would unjustly invade the self-government rights of the plurality of political communities that has always populated the political map of the world. Conversely, shrinking the scope of inclusion of affected interests to the internal ones of resident citizens would unjustly ignore the pervasive interconnectedness of human affairs across political boundaries in the present world.

The three inclusion principles have mostly been considered rivals because they tend to be linked to different conceptions of democracy. AAI is rooted in utilitarian and public choice views of democracy according to which its legitimacy and advantage over alternative forms of political rule lie in its capacity to maximize the satisfaction of political preferences and to resolve collective action dilemmas in the production of public goods. ASC is rooted in a liberal conception of democracy as the system of political rule that is most likely to guarantee fundamental rights. And the stakeholder conception is rooted in a liberal-republican conception of democracy that regards it as the always imperfect but closest possible approximation of collective self-rule.33 While there are clearly tensions between these views of democracy, they are not irreconcilable. On the contrary, it is plausible that – as in the story about the blind men describing an elephant based on the parts that they touch – from each of these perspectives an important aspect of democracy can be perceived, but not the whole.

Synthesis requires abandoning holistic claims raised in defence of each of the three inclusion principles. The elephant is not a snake, a spear or a pillar; it does not even consist of any of these parts. The three inclusion principles can be reconciled with each other only if they are stated in such a way that the claims they support can be analytically separated. The crucial step is to drop the idea that AAI and ASC serve the purpose of determining membership in a demos. The following versions of the three principles can be endorsed simultaneously and without contradiction:

The fact that these versions of AAI, ASC and ACS are compatible with each other does not mean that there are no conflicts between them. I will conclude section 3 by demonstrating how democratic inclusion norms generate tensions that cannot be resolved through theoretical reflection alone.

The first step follows from the solution of the democratic boundary paradox that I have proposed: in order to provide democratic legitimacy for policies, governments and polities, democratic inclusion must be sequenced. Policies cannot be democratically legitimate if they have not been adopted or supported by a democratically legitimate government. And a democratically legitimate government must in turn have been authorized by an inclusive demos.

In a second step we realize that the combination of the three principles as stated above does not yet provide a full account of the democratic process. The sequence from polity to government to policies reveals significant gaps that have to be filled with partially conflicting interpretations of the three principles. A political community must first constitute itself as a legitimate polity through including all citizenship stakeholders and adopting a constitution under which it can authorize a government through democratic elections before this government can provide equal protection to all subjected to its laws. A government authorized by citizens through democratic elections must also be accountable to those citizens. Moving from ACS to ASC requires, however, constraints on the popular mandate of democratic governments through constitutional and judicial protection of the rights of all subjected to the laws and not only of those citizens who can hold it electorally accountable.

A similar gap emerges in the transition from ASC to AAI. The democratic legitimacy of policy decisions depends on a prior process of debates in a wider public sphere and deliberation in decision-making bodies that track the interests of all who would be affected by the policy options on a political agenda. The institutions of parliamentary democracy and independent courts are generally not sufficient for this task. Yet there is a tension between giving voice to special interests and securing equal protection to all individuals within the jurisdiction. The tension becomes even stronger when the interests of non-citizen non-residents are included who normally do not have either a vote in democratic elections or access to the national courts.

Table 1: Conditions for democratic legitimacy

Table 1 presents a more systematic account of the complementarity and tensions between the three inclusion principles. The diagonal in this table shows the specific domains of inclusion to which each of the three principles applies. If we just focus on these cells, then there do not seem to be any conflicts.

When reading the table horizontally, that is, row after row, we see that the three principles not only have different scope but also different capacity to provide legitimacy. Three cells in the upper left corner remain empty. The underlying proposition is that AAI is a single purpose principle, whereas ASC and ACS serve dual and triple purposes respectively. In my view, AAI is uniquely suited to provide legitimacy for policy responses on issues with dispersed and border-crossing impact. For this reason, it has been invoked mostly by theories of global or transnational governance. Yet governance is not the same thing as government. The latter needs stable and bounded jurisdiction, mostly of a territorial kind, and AAI is therefore not capable of providing legitimacy for government. Even less is it a useful principle for specifying the legitimate boundaries of self-governing political communities.

By contrast, ASC does convey legitimacy to governments if they provide equal protection and opportunities for contestation for all within their jurisdiction. Yet such legitimacy would be rather meaningless if it could not be connected to norms for policy-making. Democratic governments must take decisions that go beyond the task of securing fundamental rights. To gain the kind of legitimacy that I have suggested would also be available to an enlightened autocratic government that is guided by equal concern and respect for all those included in their jurisdiction, and this means pursuing their common good. An ASC conception of democracy can therefore account for the tension between securing rights and promoting the common good that is inherent to liberal democracy and is institutionally articulated through divisions of power and mutual checks and balances between government institutions.

Stakeholder citizenship is the only principle among the three that addresses the membership question independently of a prior account of policy and government legitimacy, which is in my view the condition for resolving the democratic boundary problem. The table suggests also that ACS is the only principle that applies across all three domains by yielding specific conditions for legitimacy of governments and policies: governments are authorized by citizens and therefore accountable to them. Citizens and only citizens have the power to elect and change democratic governments. This implies also a standard for democratic policies: they must not only serve the common good of all who are subjected to the law, but they must also track the will of the people with regard to the law. There is thus again a potential for tension between the interpretations of ACS when applied to the three domains. As discussed in section 3.3, the citizenry and the demos cannot be co-extensive, but exclusions are inherently problematic. After overcoming formal exclusions on grounds of gender, race, religion and class, those remaining focus on cognitive capacity and desert (when criminal offenders are disenfranchised) and they remain problematic, even regarding age, for which thresholds and alternative modes of participation continue to be contested. Another potential tension exists between tracking the will of the people, which is required under the collective self-government aspect of ACS, and the inclusion of all citizenship stakeholders, which is required under the individual autonomy and well-being aspect. Inclusion of all stakeholders and exclusion of non-stakeholders must be translated into admission policies that are among the most hotly contested political issues in contemporary democracies.

Reading the table horizontally, the incompleteness of AAI and ASC and the inherent tensions within ASC and ACS suggest that we indeed cannot accept any of the three principles as a free-standing and full account of democratic legitimacy. Instead, we must try to combine them, which means reading the table vertically, that is, column after column. Yet doing so reveals even more tensions.

Since ACS is the only relevant principle for determining membership, interpretations of AAI that call for including as citizens all whose interests are affected by policy decisions should be firmly rejected. The case of membership inclusion on the basis of ASC is stronger, but subjection to the laws is neither a necessary nor a sufficient condition for citizenship claims in contexts such as colonial and minority oppression. And whenever it is sufficient, as it is in the case of long-term resident immigrants, the normative case for inclusion as an individual option rather than a duty can be stated better on grounds of stakeholdership.

While the absence of plausible alternative answers to the membership question means that there are no tensions in the first row of the table, there are significant ones in rows 2 and 3. A democratic government must show equal respect and concern for all within its jurisdiction, but its democratic mandate is derived from those who are citizens of the polity. Under which conditions is there no discrepancy between the answers to the two corresponding questions: Whose fundamental rights must a democratic government protect? To whom must democratic governments be accountable? First, if there is no migration between societies and, second, if citizenship is residence-based, as I claim it ought to be in municipalities. Under either of these two conditions, the answer to both questions will be: all persons residing in the jurisdiction. If in static societies all residents and only residents are citizens, this cannot determine the citizenship status and claims of immigrants and emigrants in relatively sedentary societies exposed to significant migration flows. The second condition needs some broader normative reflection that I will present in section 4 of this essay. Why should voting rights and thus membership in the demos not be derived from residence in all polities that experience significant migration across their borders? My answer will be that principles for determining citizenship and the franchise vary with the nature of the political community and that the conditions that make it appropriate for local citizenship to be based on residence are absent in the international state system.

Table 1 indicates the greatest potential for tensions in the domain of democratic policy decisions. Tracking all affected interests will often conflict with giving priority to the common good of all included in the jurisdiction. And the common good of all may conflict in turn with the will of the people as expressed through democratic procedures. These tensions cannot be simply resolved through theoretical fiat by giving priority to one or the other inclusion principle. They must instead be worked through in democratic processes of contestation and deliberation within each polity and “democratic iterations” (Benhabib 2004) across the plurality of polities. With regard to policies that aim at securing fundamental rights and protecting interests of those inside or outside the jurisdiction, the will of the people must sometimes be bound through constitutional guarantees that cannot be overruled by popular mandate (i.e. through referendum or ordinary parliamentary legislation). Yet constitutional constraints on the popular will must ultimately be accepted by the people as expressing its constitutional identity and thus its higher order will to protect such rights and include such interests (Tully 2002). In other words, resolving the tensions will only be possible if a democratic people is capable of integrating the interests of externally affected outsiders and of non-citizens subjected to its laws into its own political will.

The outcome of these reflections is not a simple and elegant theory of democratic legitimacy and inclusion. My attempted synthesis has not revealed a meta-principle from which all three inclusion principles could be derived. But this does not mean that the attempt has failed. We can read the result of this analysis in a different way: a normatively attractive conception of democracy must be pluralistic not only in the two senses that I have initially suggested but in a third sense as well. In section 2 I proposed that democracy presupposes an internal diversity of interests, ideas and identities as well as an environment populated by a plurality of bounded democratic polities. We can now add that democracies ought to accept also a plurality of inclusion principles that apply in different ways to their policies, governments and political communities. The task of political theory is to articulate the tensions between these principles rather than to resolve them. The task of coping with them in real-world contexts is a practical one. It needs to be addressed through intelligent institutional design and prudent political action.

4. Constellations and membership character of polities

Normative theories of political boundaries often lead to prescriptions that jar with our well-considered moral and political intuitions. In some cases, such as Goodin's argument that only a global demos can be legitimate (Goodin 2007), these conclusions follow from applying the wrong inclusion principles to the membership domain. In other cases, such as de Schutter's and Ypi's defence of mandatory naturalization of immigrants (de Schutter and Ypi 2015), they follow also from considering democratic polities in isolation from each other and paying insufficient attention to specific contexts.

Our discussion of democratic inclusion principles has so far been rather abstract in the sense of being decontextualized. The circumstances and territorial contexts of democracy that I outlined in section 2 apply to all contemporary democracies. None of the current politically contested boundary issues I listed in the introduction can be properly addressed as long as we stay at this level of generality and do not pay closer attention to contexts.

In this essay, I refrain from discussing particular cases, except by way of illustrating more general problems. What I try instead is to theorize contexts by proposing a general typology of political communities and relations between them. The basic intuition behind this move is that norms of inclusion across political boundaries must be specified for constellations of polities (Bauböck 2010) rather than for a single polity considered in isolation. This follows from my initial assumption that the coexistence of a plurality of bounded polities forms the background that makes democratic inclusion both possible and necessary. In other words, a theory of democratic inclusion must always account for the presence of other polities, in which those whose inclusion claims are considered may have been previously members, may become future members or may have simultaneous membership claims.

A second consideration is that the constellation that we call the international state system is a historically very peculiar context and even today not the only relevant one for democratic inclusion claims. As a normative background structure for membership claims, the contemporary international state system is characterized by non-overlapping territorial jurisdictions, the absence of any inhabitable terra nullius and legal equality between independent states. There are other constellations characterized by vertically nested polities that are neither independent of each other nor necessarily equal among each other in their legal status and powers of self-government (Bauböck 2006). I will briefly discuss the general properties of such constellations in the next section before considering how inclusion norms apply in such contexts.

My discussion will from now on focus on membership inclusion, setting aside the important task of further specifying contexts of application for the AAI and ASC principles.34 The two aspects of my general definition of stakeholder citizenship give rise to two questions that can guide us when considering how to apply it to laws and policies in real-world contexts:

1. Why and how do the autonomy and well-being of an individual or group depend on collective self-government and flourishing of a particular political community, given the way the latter is embedded in a larger constellation of polities?
2. What are the conditions under which a particular polity embedded in a constellation can become or remain self-governing without encroaching on equally legitimate self-government claims of other political communities, and what norms of membership inclusion correspond to the purpose of self-government under these conditions?

4.1 Multilevel democracy

In contemporary democratic states there is a strong tendency to introduce or to strengthen the powers of existing levels of government below the national one and a weak tendency to create new institutions for joint government at a supranational level. These historical trends add vertical depth to the dominant global structure of the international state system described above. Instead of a two-dimensional political map of the world, the global constellation of democratic polities can be visualized as a three-dimensional image with a dominant flat horizon formed by sovereign states, a few hills where supranational polities are erected on a terrain encompassing several member states, and a deep structure of territorial subdivisions of states in the foreground.

Normative theories of democratic inclusion have been mostly written from a single-polity perspective. They consider how individuals relate to a political community or authority that is the addressee of an inclusion claim and they treat polities as if they were all separate from each other and similar to each other. This approach reflects a methodological statism that assumes the international state system as a quasi-natural background. Even for an analysis of inclusion claims in democratic states, a single-polity perspective is flawed because it fails to consider how such claims emerge only once individuals are involved in relations with several states (e.g. through migration or a history of shifting borders). A normative theory of citizenship in the international state system must therefore distribute inclusion responsibilities between states and appropriate norms will generally depend both on the relation of individuals to states and of states to each other.

The flaws of a single-polity perspective become even more obvious once we consider inclusion claims in substate polities (self-governing regions or municipalities) and suprastate polities (unions of states) (Maas 2013). Insofar as these substate and suprastate polities are governed democratically, they must determine who their citizens are just as states have to do. Yet they clearly cannot do so without regard for the membership norms of the states in which they are embedded or of which they are composed. I cannot make sense of claims to inclusion in the city of Florence, the region of Tuscany or the European Union without describing first the different nature of these polities and their relations with the Italian state.

The relations that I will consider are those of states to other states, and those of municipalities, substate regions and unions of states to their “parent states”. There are other constellations that could be added, such as those connecting transborder regions or global cities. These are more complex since they involve several states as well as several substate polities. I will leave them aside in this short general discussion.

Those who support democracy because of its unique ability to legitimize political authority should not merely welcome its horizontal spread through democratic transition and consolidation in previously non-democratic states, but also its vertical deepening through democratization of substate and suprastate polities. The traditional argument in favour of multilevel government is subsidiarity, a principle that has its origin in nineteenth-century Catholic social doctrine but has now been enshrined also in the European Union Treaties (TEU Article 5). Subsidiarity is generally interpreted as requiring that in a system of multilevel government decisions should be taken by the lowest level authority that has the competence to take them and can implement them efficiently. Subsidiarity defined in this way is of limited use for deepening democracy vertically. First, the principle applies to government output rather than to democratic input. Second, it applies within an already established system of multiple tiers of government but does not require that new layers should be created where they are absent. Third, the principle favours decentralization of competences within any hierarchy of government levels independently of the specificity of polities at different levels and of relations between them. Finally, subsidiarity is a rule of priority rather than complementarity, which makes it unfit as a principle for comprehensive inclusion in multilevel democracy. It does not make sense to say that individuals should be included as citizens primarily at the local rather than at the national or European level; they must be included simultaneously in all of these polities, albeit in different ways.

The republican principle of non-domination provides better normative support for the vertical deepening of democracy.35 This argument relies, first, on the empirical premise that executive government and public administrations are structured territorially so that certain decisions are taken or implemented by local or regional level authorities. The reason for this is that in modern states government tasks are so comprehensive and government control over the population is so pervasive that a completely centralized machinery would be extremely inefficient and would soon crash under its own weight. To put it the other way round: in a libertarian night-watchmen state that only provides internal and external security and that can do so efficiently through centrally organized police and army forces, there would be no strong demand for multilevel democracy.

The next step consists in identifying the risks of domination in a single-layered system of democracy in states with a comprehensive and territorially structured public administration. The first risk is that at substate level local and regional populations would be dominated by national majorities. Since the national regime is democratic, all citizens have equal control over the institutions of national government that are in charge of administrating substate territories. But the populations living there would have just a fraction of that control that is proportional to their share in the national demos. The government of each of these territories would thus be accountable to a much larger demos in which each substate population forms a permanent minority. Substate territorial populations would thus be exposed to arbitrary exercises of power on behalf of national majorities, which is exactly what domination means according to neo-Roman republican theory. In order to avoid such domination, territorially devolved executive power and public administration must be put under the control of democratic institutions representing the citizens of these territories. This argument provides not only a reason for preserving local or regional self-government, which often pre-dates the emergence of the modern state, but also for introducing it in unitary and highly centralized states.

My non-domination argument for substate democracy seems to correspond to historical trends. By and large, democratic self-government has become stronger at substate levels. While the strength of regional level self-government varies greatly between federal and unitary states, regional elections and parliaments have become more common even in historically very centralized states, such as France and the Scandinavian countries.36 There is less variation with regard to local democracy: locally elected mayors and/or assemblies with policy-making powers are a nearly universal feature of consolidated democratic states.

The second risk of domination is much more contextual. Where states have been formed by uniting historically self-governing territories, such as provinces or colonies of empires, or where they include territorially concentrated national and indigenous minorities, substate self-government becomes a constitutive feature of the larger polity. In these contexts, the legitimacy of democratic government at the level of the wider state depends on preserving or introducing territorial autonomy at substate levels, with the consequence that these substate territories have potential claims to independence if their self-government rights are severely and persistently violated (Bauböck 2000; Patten 2014: ch. 7).

This risk of domination by national majorities is quite obvious in the case of distinct territorial minorities that lack powers of self-government. In mono-national federations such as Germany or the United States, where regional identities are not in tension with a shared national identity, the risk of domination may be less apparent. Yet a federal demos consists in any case of the citizens of the federation as well as of the citizens of the constitutive territories. The fact that these are identical sets of individuals does not mean that citizens of the federation cannot dominate those of the constitutive territories. If all political decisions were determined exclusively by majorities among citizens of the federation, then the citizens of each constitutive territory (assuming that no single territory has more than 50 per cent of the total citizen population) would be vulnerable to arbitrary exercises of power that fail to respect their constitutional right to self-government. Since federal democracy conceptually implies multilevel government, such a scenario is impossible while upholding a federal constitution. However, we can still consider it relevant for a hypothetical change that would merge constitutive territories, weaken their powers or abolish regional level self-government altogether.37 Such a decision could not be legitimately taken by a majority of the federal demos or its representatives in a popular chamber of parliament without counting separately the votes of the affected substate territories.

Risks of domination through single level democracy emerge also at the suprastate level. Today's international state system is not a Hobbesian state of nature, but thickly populated with intergovernmental organizations and regulatory regimes. Political decisions taken within these suprastate institutions and legal norms adopted by them are seen to be democratically legitimate if they are non-coercively endorsed by governments that are themselves representative of their citizens. This assumption can be challenged from two sides. One objection is that a strict condition of unanimous consent backed by democratic support for government policies in each participating state turns the demos of each state into a veto player in international governance and prevents thus any effective policy responses to problems affecting large numbers of states. The domination is in this case exercised by individual states blocking intergovernmental policy solutions that are democratically endorsed in most other states. The converse risk is that international regulatory regimes are imposed by individual states or private actors, such as global corporations, wielding superior military or economic power, in such a way that non-compliance or opt-outs are no longer feasible options for states whose citizens prefer alternative policies. Wider forums for democratic participation and deliberation of stakeholders in intergovernmental decision-making (Bohman 2007; Macdonald 2008) provide a plausible response to either risk.

Yet, as I argued in section 3.1, this response should not be confused with the creation of a self-governing demos. A supranational demos can only emerge within a permanent union of states that pool their sovereignty to a significant degree by creating legislative institutions with independent agenda-setting powers and mechanisms for enforcing collectively binding decisions that do not rely exclusively on voluntary compliance by member state governments. Although the EU is still perceived by many observers as an intergovernmental organization, the presence of such powers makes it plausible to characterize it as a supranational polity. Within such a union of states, the risk of domination through intergovernmental unanimity requirements blocking decisions that are democratically widely endorsed and the risk of domination by powers that are able to impose decisions that are democratically not endorsed are similar to those arising in intergovernmental institutions. The response to these risks must be, however, different. It is not sufficient that the supranational government gains its legitimacy from joint legislative powers of democratically legitimate governments and provides other policy stakeholders with access to deliberative arenas and contestation opportunities. It must also derive its own legitimacy directly from the citizens of the union who exercise their powers of self-government not only at state and substate, but also at a suprastate level. There is thus an additional risk of domination within a union of states that is the converse of central governments subverting substate autonomy. If democratic citizenship remains confined to the member states’ separate demoi, the citizens of the union will be dominated by the citizens of the member states through the exercise of supranational power that escapes the control of the wider demos of which they are jointly equal members. Consider as an example the abysmal failure of the EU to adopt and implement an appropriate burden-sharing and relocation scheme for refugees during the 2015/16 crisis. This failure was due to the decision of individual member states, with ample support from their voters, to block any such scheme. As a result, most refugees ended up in those states that had kept their borders initially open in 2015 or remained stuck in those countries that were the first points of entry on the refugee migration routes. In other words, the governments and citizens of those states that were not willing to take in refugees imposed on the citizens of the Union a morally arbitrary supranational distribution of responsibilities and burdens.

This argument for the need for democracy in a politically integrated union can be defended on grounds of both the ASC and ACS principles. From the perspective of the former, the difference between intergovernmental organizations and supranational unions is not so obvious and depends merely on the degree of executive or judicial powers to which individuals are collectively subjected. A stakeholder perspective draws a sharper line in requiring that the citizens of the union must be able to see themselves and each other as members of a transgenerational political community whose government institutions have to track the collective will of the citizenry. This more demanding republican conception of democratic community cannot apply to international organizations and many would also contest that it applies to the European Union in its present shape. Yet the Union has already grounded its political legitimacy to a large extent on equal treatment and democratic representation of the citizens of the Union, which is why at least in Europe citizenship above the nation-state is no longer a chimera.

4.2 Birthright citizenship

The principle of including all citizenship stakeholders, which I have defended as the best solution to the democratic boundary problem, assumes a correspondence between individual and collective interests in self-government. It implies therefore the need to consider the conditions under which a polity can be self-governing over time and in relation to other polities. Since polities are of different kinds and involved in different types of relations with other polities, ACS must be further specified when applying it to distinct types of polities. For example, the principle itself does not tell us whether residence in a territorial jurisdiction is a necessary or sufficient condition for holding a stake in its self-government and having therefore a claim to citizenship. I will argue in this and the next two sections that the answer to this question depends on the “membership character” of the polity. We can define the membership character of a democratic polity empirically as a basic rule that determines who holds its citizenship. For independent states this rule is that citizenship is acquired automatically at birth and is presumptively held for the rest of one's life.

I use the expression “membership character” rather than “membership rule” for two reasons. First, membership rules operate at different levels of generality, while the character of a polity refers to its most basic features. Birthright citizenship is the membership character of all independent states, but they differ from each other in their specific mixes of the two birthright rules – by descent (ius sanguinis) or by birth in the territory (ius soli). Moving down the ladder of abstraction, we can next distinguish specific “modes of acquisition” of birthright citizenship, such as ius soli at birth and after birth, or ius sanguinis ex matre and ex patre.38 Nationality laws contain even more specific rules determining conditions for the application of these modes, such as prior legal residence of parents as a condition for ius soli in most European states that apply it to second generation immigrants. Finally, there are rules of implementation that are often spelled out in ministerial decrees or that emerge from observing informal administrative practices. The comparative study of citizenship status is concerned with examining all these types of rules and explaining their variation. A normative theory of democratic inclusion should instead provide a critical standard for assessing such rules against a principle for democratic inclusion, while not being overly prescriptive since democratic legislators must have some leeway to adopt citizenship laws that differ from those of other democracies.

Second, “membership character” suggests that we are concerned here with a constitutive feature of the polity rather than with a rule that it could adopt, abandon or modify. If a state turned automatically all its residents into citizens and deprived all non-residents of their membership, or if it derived its own citizenship from that of a supranational union, it would cease to be recognizable as an independent state. Conversely, if a substate region started to restrict local membership and voting rights to those born in its territory or to descendants of its current citizens and treated the citizens of the larger state as foreigners who have to apply for naturalization, it would in important ways act like a state and signal its desire for independence.

Automatic acquisition linked to circumstances of birth and lifelong retention by default are the core features of birthright citizenship. This does not mean that all birthright modes of acquisition are fully automatic and determine citizenship at birth. For example, children of the second generation born to two foreign parents in France acquire French citizenship through ius soli as an entitlement around the age of majority rather than at birth and they have the right to decline the offer. What it means instead is that, while their rules differ significantly for those born to non-citizen residents or non-resident citizens, all independent states have rules that lead to automatic acquisition at birth for those born to citizen parents residing in their territory. This suggests that the basic function of birthright citizenship is to secure the transgenerational reproduction of a citizenry in a way that involves neither individual nor collective political choice.

Birthright membership also does not mean that citizenship cannot be acquired in other ways. All democratic states have provisions for the naturalization of immigrants. There is a lot of variation with regard to conditions and procedures. A stakeholder criterion provides a clear normative standard for accepting some state practices while rejecting others. From this perspective, long-term resident foreigners are obviously citizenship stakeholders, which makes all criteria apart from residence suspect.39 Moreover, naturalization should be an entitlement for those who meet the residence condition rather than a discretionary decision by the authorities. Yet the crucial point here is that naturalization is never automatic or mandatory and requires an application by the individual concerned.40 Ius domicilii is no independent ground for acquiring state citizenship of the same kind as birth in the territory or descent from citizen parents; residence is only sufficient when combined with an expression of will to join the political community.41 Naturalization therefore does not challenge the membership character of states but affirms it instead: by applying for naturalization, immigrants declare their intention to join a political community based on birthright. Whether this application requirement can be justified depends on the legitimacy of birthright from a stakeholder perspective, which I will defend below.

Finally, presumptively lifelong citizenship does not mean that birthright citizenship cannot be lost. All democratic states have rules for voluntary renunciation of their citizenship and most also have rules for involuntary deprivation. Renunciation is in a way the converse of a naturalization entitlement: the right to change one's citizenship status is in both cases held by the individual and exercised through an application or declaration. While those who naturalize normally reside in the territory, those who may renounce normally have to reside abroad. Both naturalization and renunciation also involve a relation of the individual to another state; in the former case a foreign national becomes a citizen, in the latter a citizen becomes a foreign national.

Unlike voluntary naturalization and renunciation, involuntary deprivation is hard to reconcile with a birthright conception of citizenship under which the citizenry reproduces itself automatically across generations and under which naturalization attributes to new citizens the same lifelong membership as to those who acquired the status at birth. Why should states have the power to take away what they do not have the power to grant or withhold? The grounds of deprivation that can be found in nationality laws are many and they rely on distinct reasons. We may group them into five: (1) reasons related to public security concerns (e.g. treason and other forms of disloyalty or crimes against the state, military or other service for a foreign state); (2) non-compliance with conditions for retention of citizenship (e.g. a requirement to renounce a foreign citizenship); (3) flawed acquisition of citizenship (through individual fraud or administrative error or abuse); (4) derivative loss (e.g. in case of loss of citizenship by a spouse or parent); and (5) loss of genuine link (e.g. long-term residence abroad or acquisition of a foreign citizenship). Bauböck and Paskalev (2015) claim that only revocation in case of fraudulent acquisition is clearly compatible with the integrity of the birthright citizenship regime and even this ground of loss must be qualified by the avoidance of statelessness and a statute of limitation that takes into account the time of residence and of holding citizenship status. It may seem paradoxical to claim that from a citizenship stakeholder perspective even deprivation on grounds of loss of genuine link is highly suspect, but this follows from the assumption that the genuine link is one to a birthright-based polity. Long-term residence abroad or acquisition of a foreign nationality is not a sufficient indicator for a loss of genuine link if there is a presumption of lifelong membership and the individual has not herself declared the intention to renounce her status.

At the beginning of this section, I stated that birthright is the empirical core of the membership character of independent states. I have now suggested a normative argument that the integrity of birthright citizenship should also be defended against states’ powers of deprivation. This leap from empirical to normative reasoning still needs to be backed by an ethical defence of birthright that I will outline below. Yet first I need to make clear that birthright is not itself a normative principle whose application guarantees that citizenship is allocated fairly. Depending on the specific modes of acquisition, birthright can be either severely exclusionary or over-inclusive. In other words, I am suggesting that birthright is generally compatible with a citizenship stakeholder principle and without alternative for state-based polities, but that specific birthright regimes and rules must always be assessed by using this principle as a critical yardstick.

The exclusionary nature of purely ius sanguinis-based regimes, which are still prevalent in continental Europe, has been highlighted in the critical and normative literature on immigrant integration (see Dumbrava 2014). It is indeed unjustifiable that the children of immigrants who are born and raised in the territory are treated as newcomers who have to apply for naturalization in order to become full citizens. This birthright deficit can be overcome by combining ius sanguinis with ius soli. What is less commonly pointed out is that pure ius soli regimes, which are prevalent in the Americas, are similarly exclusionary with regard to the so-called generation 1.5, that is, those who immigrate after birth as minor children.42 There is no justification for asking residents who have spent their childhood and school years in the country to wait until the age of majority before they can become citizens. The solution is in this case an inclusive naturalization regime that drops age and other requirements that cannot be defended from a stakeholder perspective. Sweden, for example, applies ius soli only to foundlings and otherwise stateless children, but the parents or guardians of minor children who have spent three years in the country can get them Swedish citizenship by a simple declaration.

Ius sanguinis is also rightly criticized for being over-inclusive if it is stretched beyond the second generation born abroad. It is obvious that third generation emigrants will generally not have a sufficiently strong stake in a grandparent's country of origin to claim citizenship, unless their parents have themselves renewed their links to this country through taking up residence there. In the case of second generation return (King and Olsson 2014), the next generation of children born abroad are again second generation emigrants and qualify for citizenship based on their ties to parents who are themselves strongly linked to the country awarding the status. It is more consistent with the birthright character of national citizenship to let it expire for the distant descendants of emigrants through non-acquisition at birth instead of depriving first or second generation holders of this status on grounds of long-term residence abroad and acquisition of a foreign citizenship. It seems, however, reasonable to exclude second generation emigrants who have never resided in the country from voting rights, even if they retain a lifelong citizenship status. Since voting rights are anyhow not acquired at birth but only around the age of majority, the concern about over-inclusiveness of an external franchise can be easily taken into account by tying the external franchise to a condition of prior long-term residence in the country;43 this would include not only first generation emigrants but also second generation returnees, while the children of emigrants who have never resided in the country would never acquire the franchise instead of being deprived of a birthright status.

The potential over-inclusiveness of ius soli is more often articulated by critics who worry about “birthright tourism” and “anchor babies” in countries with unconditional territorial birthright. The anti-immigration slant of this critique, which focuses on the potential for individual abuse, should not detract from its substantively sound core. In a society exposed to significant transborder mobility, the mere fact of territorial birth alone should not be sufficient for the acquisition of lifelong membership, since this would generate a distribution of the privileges and burdens of citizenship that cannot be justified from a stakeholder perspective. Why should children who were born accidentally or intentionally in a country where their parents have no claim to membership be awarded with unconditional immigration rights denied to others who grow up under otherwise comparable circumstances? Unconditional ius soli can also be over-inclusive with regard to citizenship duties, such as a military draft, imposed on individuals who have no genuine link to the country they are supposed to defend at the risk of their lives.

Many authors have used the prevalence of ius sanguinis or ius soli in a citizenship law as an indicator for the ethnic or civic character of a nation (e.g. Brubaker 1990; Koopmans et al. 2005). This is historically dubious (Panagiotidis 2015) and empirically implausible. As demonstrated by countries like Sweden, ius sanguinis regimes are compatible with inclusive naturalization, which means that the next generation of Swedish citizens by descent will be as ethnically diverse as its general population. It is true that ethno-nationalist governments have abused the idea of descent-based membership as a pretext for claiming influence over foreign territories inhabited by their ethnic kin minorities. But it is equally true that strong ius soli conceptions have provided pretexts for denying internal self-government rights of ethnic minorities or relegating naturalized immigrants to a second class citizenship status that denies them full political rights or provides them with less protection against deprivation.44

Finally, an especially problematic form of over-inclusion emerges in regimes that do not allow for voluntary renunciation of citizenship. This is the case in most Arab states, whose nationality laws are based on ius sanguinis (Parolin 2009), but also in a significant number of Latin American states where ius soli is especially strong. The old notion of a duty of “perpetual allegiance” is an illiberal perversion of presumptively lifelong citizenship that denies that individuals’ citizenship stakes can change over the course of their lives when they develop new ties to other polities.

The upshot of this brief discussion is that unjust exclusion and over-inclusiveness are not inherent features of birthright citizenship, but the result of lopsided laws that do not combine ius soli and ius sanguinis in a balanced way or fail to provide fair opportunities for naturalization and renunciation.

Having defended the compatibility of birthright citizenship with stakeholder inclusion, I shall now consider why independent states rely predominantly on birthright for determining their citizens and whether they are ethically justified in doing so.

Birthright citizenship serves ethical purposes that alternative rules for determining membership would either be unable to meet or not fulfil equally well. In the international state system, it minimizes statelessness and assigns to states stable responsibilities for particular individuals. Birthright contributes, first, to the prevention of statelessness through the relative administrative ease of determining birthplace or descent and the automaticity of awarding citizenship when these criteria are met. Second, states have a duty under international law to readmit their own nationals, and birthright citizenship makes it more difficult for them to shun this duty. If state citizenship were instead grounded on residence or consent, states could much more easily dump their unwanted citizens on other states by forcing them into exile and depriving them of their membership status.

From a domestic political perspective, birthright citizenship serves to promote a sense of transgenerational continuity of the polity among its citizens and reduces the risk of political conflicts over membership boundaries. First, if citizens see each other as members in a transgenerational political community they are more likely to adopt a long-term orientation towards the common good and to include the interests of future generations of citizens in their policy preferences. Second, birthright citizenship depoliticizes the determination of membership and turns it into a quasi-natural social fact. It constrains opportunities for rulers to manipulate citizenship by excluding adversaries and handing out membership as a special favour to loyal supporters. Birthright citizenship prevents also that the basic composition of the citizenry is exposed to the preferences of democratic majorities. Full and equal membership of territorial or ethnic minorities is not open to question if their members are citizens by birth.

From an individual perspective, finally, birthright provides minor children with guaranteed access to a secure legal status. Ius sanguinis ensures, moreover, that they share their parents’ citizenship, which prevents families being torn apart should parents and children have different migration and residence entitlements. Lifelong citizenship status acquired at birth also ensures that children's citizenship is not affected by the death or divorce of their parents.

These arguments strengthen the ethical case for birthright citizenship given the nature of states and the basic structure of the international state system. But they do not yet address the more fundamental moral question of whether a person's circumstances of birth should determine her membership in a political community and the specific benefits and burdens that result from such membership.45

Liberals often point out that circumstances of birth are contingent facts that should not impact on a person's opportunities and rights. The first thing to note about this critique is that, from a first person perspective, where and to whom one is born are not at all contingent facts. They are instead unchosen and fundamental features of a person's identity. The normative question is whether it is morally arbitrary to use these facts as criteria for determining membership in a political community. The charge of moral arbitrariness relies on two claims: (1) newborn babies are not responsible for the circumstances of their birth and thus cannot have deserved to be born as citizens of one state rather than another; and (2) citizenship in different states entails unequal opportunities and rights.

The first of these claims relies on the premise that desert is a relevant criterion for the allocation of citizenship. This premise is not only incompatible with the citizenship stakeholder principle that I have defended, but it would also fatally undermine any conception of citizenship as an equal status in a territorially inclusive political community. The second claim is empirically undisputable and morally indeed troubling. However, once it is separated from the first claim, it becomes clear that the target of critique cannot be the automatic attribution of citizenship based on circumstances of birth. The target must instead be the inequality of opportunities and rights that different states offer to their citizens.

Inequality between states can originate from political decisions taken by legitimate governments authorized by their citizens, in which case it is unobjectionable from a pluralist democratic perspective. Or it can emerge from unequal natural and social resources (including historical conditions) for which current citizens and governments cannot be held responsible. Finally, it can result from inter-state or global forms of domination for which other states can be held responsible. Liberal theories of global justice will differ in their attitudes towards the second source of inequality, but ought to agree that the first source is normatively unproblematic while the third gives rise to claims for redistribution or redress. I do not have to enter here the vast literature on global social justice, since the conclusion for our present concern is straightforward: inequalities of rights and opportunities for the citizens of different states have to be addressed independently of whether or not citizenship is determined by birthright.

Joseph Carens's famous argument that, from a global perspective, birthright citizenship is like a feudal status (Carens 1987, 2013 ) relies on an additional empirical premise: states control immigration and preserve global inequality by preventing the citizens of poor countries from seeking better opportunities in wealthy ones. This argument raises several new questions:46 Is immigration control an inherent feature of states or can we imagine a world of states with open borders and birthright citizenship? As I will argue below, this is not a fanciful idea if we consider freedom of movement and open borders in the current European Union, whose member states are also full members of the international state system and whose citizenship is still based on birthright. Would open borders between states be an adequate remedy for inter-state inequalities of the second and third kind? While dismantling migration restrictions is likely to contribute to reducing global inequality indirectly it is probably a very ineffective remedy for the misery of the globally worst off populations who lack the social and economic resources needed for migration to wealthy destinations.

I will further explore the link between citizenship and free movement in the following sections. For now, my conclusion is that birthright is an implausible culprit for those injustices that liberal egalitarians rightly worry about.

4.3 Residential citizenship

The birthright regime that characterizes contemporary states has its historical origin in the city. A democratic conception of citizenship as membership in a self-governing political community originated in the Greek and Roman city republics and was later revived in the late medieval and early modern free cities of northern Italy, Germany and the Netherlands. These cities’ membership regimes resembled in many ways those of modern states: they had fortified borders and gates for controlling immigration, and the status of a citizen or free burgher was generally acquired through birthright or naturalization. By contrast, the borders of the empires within which most of these city states flourished were loosely controlled and frequently shifting frontiers rather than stable demarcations of jurisdiction. Empires did not have citizens but turned everyone inside their territory into subjects, while they had little leverage over those who had left and were thus often more concerned about controlling emigration than immigration.

Of course this description of boundary structures does not do justice to the endless historical varieties of city polities and empires. I use it here only to provide a contrast that helps to sharpen our perception of the historically contingent and quite exceptional nature of membership boundaries in our current world. My point is that the boundary constellation that once related free cities to empires has been turned inside out in creating the contemporary boundary constellation of democratic states and their self-governing municipalities. Today birthright citizenship is a universal and nearly unique characteristic of independent states, while open borders and residence-based membership form the boundary conditions for local self-government.

This historic reversal has come about through what the British sociologist T.H. Marshall described as the “geographic fusion” and “functional differentiation” of citizenship (Marshall 1949/1965: 79). As long as citizenship was primarily local, cities were in control of settlement in their jurisdiction, and the institutions providing for justice, political representation and social welfare of urban citizens were hardly differentiated. The economic and political transformations that gave birth to the modern capitalist economy and nation-state tore down the city walls, dismantled the local welfare regimes and established uniform civil, political and social citizenship rights at the national level provided through the separate institutions of courts, parliaments and public administrations.

Yet local citizenship did not vanish completely. As I pointed out in section 4.1, local democracy is a universal feature of democratic states. Where local councils or mayors are elected and where referendums on local policies are held, there must be a corresponding demos of citizens eligible to participate in these elections. Using this empirical test, we would arrive at the conclusion that the local citizenry is determined differently across contemporary democratic states. In most, the franchise in local elections combines the conditions of residence in the municipality with that of nationality of the encompassing state. However, this rule is not applied consistently. Under EU law, local voting rights (but not the franchise in regional or national elections) are also awarded to the nationals of other member states residing in the municipality. Portugal and Spain grant local voting rights to specific other countries on the basis of reciprocity. In the UK, these are extended to Irish and Commonwealth citizens (with no reciprocity in the latter case). Most significantly, twelve EU countries, as well as Norway, Iceland and several Swiss cantons, plus a significant number of Latin American states, disconnect the local franchise from nationality altogether.47 While these empirical observations are not conclusive for a residence-based normative conception of local citizenship, they demonstrate its empirical compatibility with a birthright-based conception of national citizenship.

The normative argument can rely on three features of local self-government in democratic states that are relevant to its membership character. First, the competences of municipalities are limited to providing public goods and services for local residents, whereas independent states perform the additional task of representing their citizens towards other states. Unlike states, municipalities do not have extraterritorial personal jurisdiction. Local voting rights are therefore very rarely extended to former residents.48 Generally, only residents thus form the local citizenry. This does not mean that local policies are necessarily territorially confined. Towns engage frequently in twinning projects across borders, and the governments of global cities are important players in transnational issues such as trade and finance, migration and environmental policies. City governments often borrow policies from cities in other countries and deliberative bodies of city legislators and executives could work out solutions to some global problems upon which states are unable to agree (Barber 2013). Unlike federal provinces striving to enhance their autonomy, cities disconnect from the international state system when propelling their soft power across borders. In Warren Magnusson's (2011) words “seeing like a city” means considering public policy issues from a non-sovereignty perspective.

Second, the right of free movement inside a state territory is not a privilege of national citizens, but has been codified as an international human right.49 Not only national citizens have an unconditional right to establish residence in a municipality, but also non-citizens who have been admitted to legal residence in the state territory. From the perspective of municipalities that provide public services to their resident population, distinctions between those born in the city, newcomers who are nationals of the state and others who are foreign nationals are arbitrary since they cannot be grounded in any democratically legitimate purpose of municipal self-government. If this is true for policy output legitimacy, there is a prima facie case that all residents ought to be treated equally also with regard to democratic input. The local citizenry should therefore exclude non-residents and include all residents.

Third, the case for shifting from a birthright-based to a residence-based conception becomes stronger the more mobile and the less sedentary is the society within the territorial jurisdiction. By creating open internal borders inside their territorial jurisdiction, states expose local municipalities to much more mobility than they are themselves exposed to. As I explained in section 2.2, cross-border migration increases as the scale of territory shrinks. Paradoxically, it is thus the dominance of birthright-based polities over other self-governing polities inside their territory that leads to the emergence of a space for an alternative residence-based citizenship at the local level.

These three reasons are supportive but not yet decisive for my normative claim, since similar arguments could be made about counties, provinces, regions and other kinds of substate territories. The fourth reason is that citizenship should be based on residence in all substate territories that are self-governing and not involved in a constitutive relationship with the encompassing state polity. I will examine citizenship in constitutive territories in section 4.3. My argument there is that in constitutive relations between polities (such as between a federal province and the federal state or between a member state and a supranational union of states), citizenship ought to be derivative, so that all citizens of the constitutive territory are also citizens of the encompassing polity. By contrast, where substate territories are not involved in a federal relation with the state, there is no plausible reason for linking citizenships between the two levels. Doing so imposes a state-based conception of citizenship on polities that are neither states nor involved in constitutive relations with states that secure their political autonomy. In current democratic countries, municipalities are too often regarded as “creatures” of the state (or – as in the U.S. – of the several states of the federation). This constitutional doctrine allows higher level polities not only to dictate the laws for the local franchise, but also to change the borders of local jurisdictions or to dismantle them altogether. As I suggested in section 4.1, the absence of local levels of self-government amounts to a structural domination of local by state citizens. This is also true if local self-government is a dependent creature of the state.

The constitutional doctrine of the “homogenous people” invoked by the German and Austrian Constitutional Courts when striking down local voting rights for non-citizens ought to be rejected for the same reason.50 The German court claimed that the German people must be identical across all levels of the German polity and therefore cannot include foreign nationals at the local level who are excluded from the franchise in national elections. The Austrian court referred to Article 1 of the Federal Constitution that proclaims that the law emanates from the people and concluded that the people must be defined as all those possessing Austrian nationality. This notion of a homogenous franchise in all elections is anyhow incompatible with membership in the European Union, which implies voting rights for non-national EU citizens in local and European Parliament elections.

While the fourth reason provides a decisive argument for rejecting a state-imposed condition of national citizenship for local voting rights, the three other reasons explain why the normative conception of local citizenship should be based on residence rather than local birthright or local self-determination. The former option would undo the geographic fusion from which national citizenship had emerged. But why should municipalities not be able to decide for themselves whether they include only resident national citizens or also non-national residents? The reason is the same as that which led us to reject democratic self-determination with regard to inclusion at the national level. Each type of polity has to rely on a basic membership rule that must not be exposed to the political will of its demos because the rule is what constitutes its legitimate demos in the first place.51

I conclude that for non-constitutive but self-governing substate polities, the only normatively defensible rule is to consider all residents and only residents as citizenship stakeholders. This argument does not apply specifically to cities. There are, after all, independent city states, such as Singapore, and federal city provinces, such as Berlin, Hamburg, Bremen and Vienna.52 And it does not apply to any administrative substate territory where there are no local governments accountable to local citizens. The argument is therefore not so much about cities and municipalities as a specific type of polity, but about the need for spaces of residence-based citizenship inside the territory of democratic states. In democratic states, citizenship should not be based on birthright or derived from birthright citizenship all the way down. Residence-based citizenship creates an alternative space for democratic self-government in which all those who live together in a territory at a certain point in time can see each other as political equals.

4.4 Derivative citizenship

I have so far discussed two polity constellations – states in relation to other states and states in relation to substate non-constitutive polities, such as municipalities, – and I have defended birthright citizenship for the former and residence-based citizenship for the latter. There is a third polity constellation that puts states in constitutive relations to regional polities at substate or suprastate levels. Citizenship in such regional polities is derivative of state citizenship. This is their basic membership character.

What I mean by derivation is a linkage between two polities through which those who are citizens of one are automatically also recognized as citizens of the other. In nested two-level polities derivation can be either downwards or upwards. In the former case, the citizenship of the encompassing polity determines that of the constituent entities; in the latter case, it is the other way round. Contemporary federal states are nearly universally characterized by downward derivation. The regional polities have no autonomous rule for determining their citizenship. The Fourteenth Amendment of the U.S. Constitution provides the general formula: “All persons born or naturalized in the United States, and subject to the jurisdiction thereof, are citizens of the United States and of the State wherein they reside.”

In confederations or unions of states, derivation is instead upwards. As already mentioned, Switzerland seems to be the only contemporary state where federal citizenship is formally derived from that of the cantons. Upward derivation is, however, the basic membership characteristic of the European Union (as well as of more rudimentary forms of supranational citizenship in Latin America): “Every national of a Member State shall be a citizen of the Union. Citizenship of the Union shall be additional to and not replace national citizenship” (TEU, Article 9).

Since the polities involved in a federal relation are not identical, derivation must be complemented with an activation rule that ensures a smooth uptake of citizenship by those who move across internal borders. This rule can only be based on residence, no matter whether derivation is downwards or upwards. Imagine that the Fourteenth Amendment had declared persons born in the U.S. to be citizens of the United States and of the State wherein they have been born. A citizen born in New Jersey who resides now in California would then remain subject to the laws of New Jersey where she could also vote in State elections, whereas in California she would be eligible only for federal citizenship protection and benefits. New Jersey and California would then relate to each other as foreign countries that have signed an international agreement providing for a common floor of rights enforced by a government in Washington, D.C.

In the EU, the situation is different since the citizenship that is activated through taking up residence in another member state is not the national one of that state, but the European one itself. All nationals of member states are formally citizens of the EU, no matter where they reside, but most of the specific rights and protections provided by EU citizenship kick in when EU nationals are involved in “cross-border situations”, which include internal migration to another member state or just having family members or conducting business there. We can again imagine a counterfactual scenario in which member state nationals would automatically acquire the citizenship of another member state where they take up residence. This would reverse the direction of derivation and transform the EU from a union of states into a federal state.

Federal states and supranational unions are thus structurally similar, since in both cases the “source citizenship” is that of an independent state and the derivative citizenship is activated through internal mobility, while the source citizenship remains unaffected (Jackson 2001). This similarity is also suggested by the very term “region” that applies both to substate political territories and to unions of states. Although in the European polity there are four levels of citizenship (local, substate regional, state and supranational), there are only three distinct polity constellations and corresponding membership rules (birthright, residence-based and derivative). In section 4.3 I described how the modern relation between states and self-governing local polities has emerged from turning the previous constellation of free cities in empires inside out. We can now add to this the further observation that supranational regional citizenship also looks like substate regional citizenship turned inside out, with the important difference that there is no historical transition from one to the other, but their simultaneous presence in the same European polity.

The term “region” is, however, unspecific insofar as it does not entail a self-governing polity with its own citizenship. Most unions of states, such as the African Union or the Association of Southeast Asian Nations, are better described as regional international organizations rather than as supranational regional polities. Some regions group together territories on either side of international borders that share cultural similarities or economic resources. In a few cases, such cross-border regions have established consultative political bodies in which delegates of provincial governments or parliaments deliberate about cooperation. Yet such cross-border regions cannot become polities in their own right without challenging the territorial integrity of states in which they are embedded. A third type of region that is not a polity is exemplified by administrative subdivisions somewhere between the local and the national level, such as the French départements or the British counties. Regional self-government at substate as well as suprastate level is a matter of degree. It exists without doubt in all genuinely federal constitutions as well as in the European Union. In unitary states, such as France or the Scandinavian countries, there are often elected regional assemblies and thus also a form of regional citizenship. Yet in these constitutions regional self-government is not qualitatively different from local self-government. As I have already suggested in the previous section, the determination of membership should then be based on residence rather than derivation. This is what we find indeed in Denmark, Sweden and Slovakia, where third country nationals can vote in both local and regional elections (Arrighi et al. 2013).

What I am concerned with here is not a typology of regional self-government (see Hooghe, Marks and Schakel 2010), but the scope of application for derivative citizenship. For this, there must be a second condition apart from institutions of regional self-government, and this is a constitutive relation with other polities through which sovereignty is dispersed and shared. Such a relation is inherent in federal constitutions, but exists also in others that are not formally federal and need not involve all parts of a state territory. Britain and Spain are not federal states, but devolution has turned Scotland, Catalonia and the Basque Country into polities whose degree of autonomy is now a matter of negotiation between regional and central governments. The same can be said about special autonomy arrangements, as in the Finnish Åland islands or the north Italian province of South Tyrol, that set certain regions apart from the rest of the state territory. A constitutive relation is one in which there is not only multilevel self-government, but also in which the encompassing polity is composed of distinct polities with whom it shares sovereignty, no matter whether this constellation has come about through “coming together” federation or “holding together” devolution (Stepan 2001). In such a relation, the territorial integrity and political legitimacy of the encompassing polity depend on constitutional safeguards for the autonomy of the polities of which it is composed. Governments involved in these relations have reciprocal duties to respect each other's spheres of autonomy and to cooperate in areas of joint government.

Derivative citizenship is an essential ingredient of this constellation because it turns all citizens of constitutive polities into citizens of the encompassing one. Each individual citizen has thus a double stake in the self-government of both polities. Such dual citizenship is different from multiple citizenship for migrants that reflects their individual ties to independent states. It is also unlike dual citizenship at local and state levels that differentiates between resident population and national citizenry. In a constitutive relation, dual citizenship serves instead to connect polities that could potentially be separate.

The isomorphism between substate and suprastate regional citizenship is certainly not perfect. For example, just like local citizenship, substate regional citizenship generally does not have an extraterritorial dimension. A few largely symbolic exceptions, such as the embassy-like representation of German provinces in Brussels or the Foreign Ministry of Quebec and its presence in Canadian embassies and consulates around the world, confirm this rule. By contrast, EU citizenship has an explicit external dimension that includes the right to diplomatic and consular protection in third countries through the representations of other member states and EU delegations (TFEU, Article 20(2)c). Whether EU citizens residing in third countries enjoy also external voting rights for European Parliament elections depends, however, somewhat incoherently, still on national legislation of their country of nationality (Arrighi et al. 2013).

Such differences between regional citizenship within states, on the one hand, and EU citizenship, on the other hand, emerge quite naturally from the fact that the latter shares the global stage of international relations with independent states, including those of which it is composed. This raises the larger question about the transformative potential of supranational citizenship for the global political order. Cosmopolitan optimists have sometimes regarded the EU as an approximation of, or vehicle for, a Kantian world federation of democracies, while realists tend to see it as a singular European experiment that is essentially still an intergovernmental organization and inherently limited in geographic scope. I want to sketch a position between these two extremes that does not rely on geopolitical considerations but considers instead the internal constitutional dynamics of a regional union of democratic states.

The European Union differs from federal and confederal polities because of a combination of four constitutive principles: (1) a common citizenship with direct elections and representation at union level; (2) the dominance of member states in constitutional law, entrenched through a unanimity requirement for EU treaty change; (3) partial opt-out opportunities from union policies and full exit options for member states, spelled out in the Lisbon Treaty (TEU, Article 50); and (4) deeper political integration as a goal, expressed in the ever-closer-union commitment in the preamble of the Treaty of Rome.53 These are not merely principles that characterize the current constitutional arrangements of the EU; they articulate a general model of political community between the state and the global level.

In stark contrast with federal or confederal constitutions, the four principles do not aim for a stable equilibrium. While the unanimity requirement for constitutional change risks blocking reform, the exit opportunity and ever-closer-union principles introduce a dynamic vision of a polity whose composition and degree of integration change over time. The tensions between these static and dynamic elements keep the historic possibilities of disintegration or transformation into a federal state alive, but neither of these possible outcomes can be a legitimate goal pursued by member states under the current constitution.

Although the EU has grown continuously from 1957 to the 2013 accession of Croatia, I have deliberately not included enlargement among its constitutive features. The political reunification of Europe after the Cold War was the second great historical mission of the EU after securing permanent peace between the adversaries of two world wars. Yet a goal of continuous enlargement is particularly hard to square with a unanimity requirement in treaty legislation and a commitment to ever-closer union. The reason why, until the Brexit vote, the EU has been capable of simultaneously deepening its political integration and expanding to twenty-eight member states is that each new accession was conditional upon accepting the acquis that entrenched the current state of integration. Had the Union started from an initial number of twenty-eight states, it would have probably remained an international organization focused on common markets. Supranational citizenship and democracy would have been a very unlikely achievement under this condition (Lacey and Bauböck 2017). The tradeoff between enlargement and political integration is ineluctable and increasingly obvious in current attempts to overcome the divisions between debtor and creditor states.

The upshot of these considerations is that supranational democratic polities such as the EU can be successful in terms of their declared goals only if they remain limited to regional unions. If the historic experiment of the EU succeeds, it will neither result in a new regional superstate nor in an ever-expanding union with claims to global hegemony. Instead, it will be imitated in other parts of the world, thus reinforcing the pluralistic circumstances of democracy by populating the world with a novel type of polity – democratic regional unions of states.

The ethical purposes of derivative citizenship are not universal ones, unlike those of birthright and residence-based citizenship. Not all democracies need to adopt federal constitutions or introduce autonomy arrangements for some parts of their territory. And not all democracies need to join democratic regional unions. The imperatives for such solutions arise in specific historical and geographic contexts. But where they arise, they do have considerable normative force. Moreover, from a global perspective we can also see how the dispersal and pooling of sovereignty at substate and suprastate levels reduces the risk of political domination within states and enhances opportunities for democratic self-government beyond the state.

5. Conclusions: maintaining self-government under conditions of polity plurality

In section 2 of this essay, I claimed that a plurality of polities of different kinds should be accepted as a basic feature of the circumstances of democracy. I suggested that the three democratic principles of including all affected interests, all subjected to coercion and all citizenship stakeholders are equally valid but address three different questions: Whose interests should be represented in policy decisions? Who should be protected by government power and have the right to contest it? Who should be a citizen of a self-governing political community? At the end of section 3, I acknowledged that this correspondence between inclusion principles and democratic domains is not perfect, but insisted that the membership question can only be resolved within a democratic stakeholder conception of citizenship.

In section 4, I examined how a principle of stakeholder inclusion applies to polities of different types. I suggested that this question cannot be answered by considering polities in isolation from each other. Stakeholder claims to inclusion refer to a relation between the conditions for individual and collective self-government. The conditions for collective self-government depend in turn on the relation of a polity with other polities. I examined three different types of such polity constellations – relations between independent states in the international state system, between states and local self-government, and between constitutive polities and unions of polities – and identified birthright, residential and derivative citizenship as the corresponding basic rules for determining individual membership.

I will conclude in a similar way as I ended section 3, with an attempt to synthesize and to point out some of the complexities that emerge from combining the three membership regimes. This attempt is guided by an old republican question. Republicanism is a political philosophy that gives equal weight to individual and collective freedom and connects the two. Under the influence of modern liberalism, contemporary republican theorists have shifted the emphasis towards individual freedom from domination and have tended to neglect collective freedom, while the classic republicanism of Machiavelli or Rousseau was biased in the opposite direction and primarily concerned with the conditions for collective freedom. Machiavelli, in particular, was obsessed with “mantenere lo stato” – preserving the conditions under which republican government would be able to endure. In his view, this required permanent vigilance against internal enemies of the republic (the nobles) as well as external enemies (rival polities). The threats of internal corruption and external domination have not of course vanished. But the great transformations that have created the modern state and the international system have raised new challenges for maintaining self-government that are not reflected in the classic texts. The membership regimes that I have outlined in the previous sections can be best understood and defended when we read them as partial answers to three such challenges.

The first is the challenge of how to maintain the continuity of a self-governing polity across generations. This is not itself a modern question, but it has acquired a new urgency for several reasons. One is the vanishing of quasi-natural boundaries of political community based on religion, race or ethnicity that made it easy to distinguish natives from foreigners. These markers have lost both empirical salience and normative legitimacy. Another reason is the increasing awareness that policies have long-term effects and need to factor in the interests of future generations in sustainable environments or welfare regimes. Birthright citizenship is certainly not a sufficient answer to this challenge, but it may be a necessary one once we consider the alternative of grounding all political memberships on residence. In a diverse society, both ius soli and ius sanguinis remove the divisive question of who has a claim to full citizenship from the political agenda and make citizens aware that they form part of a long chain of generations who are all unconditionally included in the political community.

The second challenge is how to maintain territorial self-government in contexts of territorial mobility and migration. I have initially assumed relative sedentariness as a context for democratic inclusion and it is important to emphasize the qualifier “relative”. Freedom of movement is an essential aspect of individual freedom that liberals and republicans care about. Yet how can there be stable democratic self-government if large numbers of native-born live abroad and large numbers of foreign-born join the resident population? In such contexts membership must be determined also on the basis of residence. In contemporary democracies this happens in two different ways. The initial allocation of citizenship through birthright is corrected through residence-based naturalization opportunities and absence-based renunciation opportunities. And inside the territory of democratic states, municipalities have open borders for mobility and accept all residents as local citizens. The difference between national level naturalization and local level ius domicilii can only be maintained as long as national societies are generally sedentary, in the sense that most mobility remains contained within state borders and a majority of birthright citizens have permanent residence in the territory. If this condition changed – and not just in a single polity but as a general feature of the international state system –, then birthright regimes would lose their legitimacy, with the problematic consequence that citizens would have a much weaker sense of transgenerational continuity.

The third challenge is how to maintain self-government in contexts where political communities are bound together – by history or choice – into larger unions. Federal democracy with multilevel citizenship was invented by the American Revolution, which has provided a template for many other federal states around the world. In our time, the European Union has developed a model for a democratic union of independent member states that has so far triggered only timid imitations. Maintaining multilevel self-government within nested polities each of which could potentially be independent requires an automatic link between their memberships through downwards or upwards derivation. Such federal and union citizenship turns individuals into multiple stakeholders that do not only have individual inclusion claims but also an interest in the maintenance of the union itself. The link between constitutive and federal or union citizenship is certainly not sufficient for resisting centripetal as well as centrifugal forces. But it should contribute to stabilizing such arrangements by articulating the ideas that the self-government of each level depends on that of the other and that every citizen has a stake in democracy at both levels.

Although there is a strong correspondence between the three membership norms and the three challenges for self-government, this is not a set of unique pairs. In other words, each of the three challenges can be seen through the prism of each of the three membership norms. Table 2 illustrates this idea. What emerges from considering all these combinations is that the three membership norms ought to be seen as complementing each other, just as I suggested in section 3.4 with regard to the three democratic inclusion principles.

Table 2: Membership norms and political challenges for democratic polities

Consider first the second cell in the first column. How can a sense of transgenerational continuity be maintained also in polities with residence-based citizenship and very high levels of mobility? In superdiverse cities (Vertovec 2007), where being native is a minority identity and large numbers are only temporarily present, people will still be aware that there are government institutions whose laws apply to them and that these institutions have been there before their arrival and will continue to rule the city after their departure. If they can also participate politically in the making of these laws, this will strengthen their sense of being stakeholders in the self-government of the city. The political time horizon of mobile urban citizens is still likely to be much shorter compared with the horizons of those who take a decision to naturalize. This could pose a major problem for political integration if cities were like independent states. It is the local polity's embeddedness in national birthright regimes that secures stability and legitimacy for the institutions of self-government also in superdiverse and hypermobile cities.

Moving down to the last cell in the first column, continuity is generally more fragile in federations and unions with derivative citizenship than in unitary states. The threat of breakup might linger on as a historical memory of past conflicts or be a present fear that undermines trust and the willingness to collaborate. Under such conditions, secession or partition into independent states, as well as merger into a unitary democratic polity, imply the discontinuity of self-government of either the encompassing or the constitutive polities. Continuity can be achieved through a constitution that entrenches self-government rights at both levels and divides powers between them in a way that minimizes permanent conflict.

Let us now look at the second column of the table and consider how birthright-based and derivative membership can cope with migration in the sense of border-crossing mobility. Although all independent states claim the right to control immigration, their birthright citizenship creates significant spaces for free movement. Birthright citizenship entails a right to unconditional residence and readmission. Since it is not lost through emigration and can be passed on to at least the second generation born abroad it creates a potential for “return migration”. Moreover, gender neutral ius sanguinis and the combined effects of ius soli and ius sanguinis have led to a proliferation of multiple nationality, which has been further enhanced through a global trend in both sending and receiving states towards accepting that a previous citizenship can be retained in case of naturalization. Multiple citizenship, however acquired, creates unconditional free movement rights between several states. Finally, international freedom of movement could be extended much further on the basis of a birthright conception, if states accepted a duty to promote their own citizens’ opportunities for migration by concluding free movement agreements with other states on a basis of reciprocity (Bauböck 2009a).

In federal states, derivative citizenship in constitutive polities does not add much to the general right of free internal movement in state territories that has become a universal human right. If anything, the autonomy of constitutive polities may sometimes justify restrictions of free movement rights for the sake of protecting vulnerable linguistic or indigenous minorities or provincial welfare regimes. By contrast, in unions of independent states derivative citizenship at the union level is a major source of free movement. In the EU it has not only led to very extensive mobility rights attached to Union citizenship, but also to a general abolishing of border controls within the Schengen zone. Moreover, EU enlargement has transformed millions from third country nationals subjected to tight immigration control into potential free movers protected by strong non-discrimination rights. Precisely because of its focus on free movement, EU citizenship may however have indirectly contributed to creating new political cleavages between mobile Europeans and sedentary nationals who do not see themselves as European stakeholders.

The third column of the table considers, finally, the capacity of birthright- and residence-based membership regimes to promote or preserve structural union between polities. The answer is generally negative. Birthright citizenship cannot be replicated across levels. If both levels have it, they are de facto independent states; if one level has it without a derivative citizenship for the other, then the latter polity is not self-governing and there is thus no union. By contrast, citizenship can be coherently residence-based at all levels, but then nested polities will be merely similarly exposed to inflows and outflows of their temporary citizens, without being linked to each other in a citizenship union.

The need for a robust derivative citizenship is especially strong in territorially divided societies. As I explained in section 4.2, the quasi-natural equality among birthright citizens protects racial, religious and indigenous minorities that had historically been excluded from full citizenship. At the same time, birthright inclusion in a supposedly homogenous political community can undermine minority self-government unless the latter is secured through constitutional autonomy arrangements and derivative citizenship. For fair inclusion of minorities that do not have a history and claim to territorial self-government the challenge is to re-imagine the larger polity as a multicultural, multiracial and multireligious one in which birthright does not need the support of thicker conceptions of national identity.

Residence-based citizenship combined with free movement is often perceived as a threat rather than a benefit by national and indigenous minorities that fear being overwhelmed by majority citizens and immigrants settling in their territory. Replacing derivation with ius domicilii is therefore not conducive to maintaining union in multinational democracies. However, there are also cases where national minorities (such as the Catalans, Basques or Scots) have strategically adopted an implicitly residence-based conception of regional citizenship as a device for stemming demographic decline and opposing restrictive immigration and immigrant integration policies pursued by central governments.

If these considerations seem rather inconclusive, this is for two reasons. The first is that they reflect on a messy reality in which polities of different types do not always respect each other's legitimate self-government claims, enjoy very unequal powers and are often locked in conflict with each other. This is a reality that we already know very well from the contrast between the norm of equal sovereignty of states and de facto dominance and inequality of power in the international state system. Just as the norm of equal sovereignty remains valid and operative for purposes of international law, despite a contrasting reality, we should think of the norms of democratic inclusion and self-government in the more complex environment populated by multiple types of polities as having validity independently of the distortions that we find in the contemporary world.

The second reason is that the commitment to pluralism that has guided my analysis in this essay is not limited to the arguments that an internal diversity of interests, ideas and identities and an external plurality of different types of polities form the circumstances of democracy in the political world. At the end of section 3, I added a third pluralism of democratic inclusion principles. Going one step further, I am also inclined to accept Isaiah Berlin's view that there is an irreducible plurality of values in the moral world (Berlin 1991). The values of individual freedom and collective self-government cannot always be fully and simultaneously realized because there is tension between them that occasionally results in deep democratic dilemmas. This is no reason to favour one at the expense of the other. The principles of democratic inclusion guide us when we try to resolve this tension.