4 - Political Parties in Constitutional Theory

Summary

In this chapter, I first argue that the key function that parties perform, when functioning as they ought to function, is to facilitate a mutually responsive relationship between public policy and popular opinion by acting as an intermediary between a state and its people. They perform this intermediary function in a unique manner, because of their bi-directionality and their plenary character. When they perform this function effectively, political parties significantly reduce four key information and transaction costs that would otherwise make democratic governance impossible: political participation costs, voters’ information costs, policy packaging costs, and ally prediction costs. I further identify four normative principles in relation to political parties: viz, the purposive autonomy principle, the party system optimality principle, the party-state separation principle, and the ‘anti-faction principle. These political principles are drawn from the value of democracy itself. As such, they should – alongside other relevant political and constitutional norms – inform fundamental constitutional design choices.

4.1 Introduction

Political parties appear to be in crisis. Lazy clichés in popular culture routinely stereotype politicians and parties as a group to be self-serving, elitist, and corrupt. The recent wave of democratic deconsolidation in several established democracies has been accompanied – perhaps caused – by the collapse, authoritarian takeover, or external capture of mainstream political parties, the partisan capture of state institutions, and a rise in hyper-nationalistic and exclusionary partisan rhetoric.¹ This chapter forms part of a larger ongoing project in defence of parties, politicians, and politics, one that is examining relatively ignored non-judicial phenomena in constitutional scholarship, such as incremental party-state fusion,² rights and powers of opposition parties,³ second chambers and semi-parliamentarism,⁴ political directives as constitutional norms,⁵ anti-plutocratic constitutional norms,⁶ and the so-called fourth-branch guarantor institutions.⁷

Political parties and party systems have long been the central institution for analysis in political science. A significant part of the political science literature is taxonomical, based on relationship between parties, their leaders, their members, the electorate, the state, and electoral systems.⁸ This literature has taught us about cadre and mass-based parties,⁹ catch-all parties,¹⁰ electoral professional parties,¹¹ cartel parties,¹² and market-oriented parties.¹³ Some of this literature has also looked at the relationship between party systems and certain institutional arrangements – such as Duverger’s law that a first-past-the-post system is likely to result in a two-party system,¹⁴ and the correlation of strong and weak parties with parliamentary and presidential systems respectively.¹⁵ The first set of scholarship shows just how dynamic, adaptable, and indispensable parties have been to numerous challenges democracy has thrown their way. The latter set shows – albeit at a very high degree of generality and typically in a monocausal fashion – how parties respond not only to the broader politico-social and economic developments but also to certain features of constitutional design. Other disciplines that have examined parties closely, although not to the same degree as political science, include political history¹⁶ and political theory.¹⁷

While political parties have long been a central object of study in political science, Anglophone constitutional law and theory scholars have, until recently, largely ignored this key democratic institution.¹⁸ The little attention constitutional scholars have given to parties either concern the jurisdictionally specific legal regulation of parties and elections¹⁹ or (rarely) the impact of electoral or party systems on specific policy outcomes.²⁰ Comparative constitutional scholars and constitutional theorists have largely been silent due to the influence of the American and the British constitutional traditions on the field, which, unlike their European continental counterparts, are largely silent on political parties. This silence is mainly a feature of big-C constitutional codes in the Anglophone world. Small-c constitutional statutes, conventions, and judicial precedents in these states do, admittedly, engage extensively with political parties.²¹ But the large-C textual silence is nonetheless indicative of the low level of salience this key constitutional institution has been given, both in constitutional practice and constitutional scholarship. Madison, echoing Rousseau,²² was famously hostile to the emergence of political parties as ‘actuated by some common impulse of passion, or of interest, adverse to the rights of other citizens, or to the permanent and aggregate interests of the community’.²³ That he would go on to found one of the two first political parties in the United States is, of course, another matter.²⁴ Contrast Madison’s ideological hostility to parties with Burke’s more affirmative opinion that ‘a party is a body of men united for promoting by their joint endeavours the national interest upon some particular principle in which they are all agreed’.²⁵ Sadly, this Burkean optimism about parties fails to inform constitutional theory despite the British influence on the field because structural features of the organic constitutional development in the United Kingdom never produced a big-C constitutional code.

Big-C codes typically design key state institutions in a democracy. Parcelling off considerations about political parties to small-c statutes and conventions has the effect that the shape of the party system becomes an afterthought, left to be regulated by small-c statutes while taking the design of key state institutions as a given. As this chapter argues, however, bringing parties to the forefront of the constitutional imagination has very important implications for how we ought to think of fundamental institutions and offices of the state. Furthermore, big-C constitutional change tends to require the buy-in of opposition parties, whereas small-c changes can usually be made by the ruling party/coalition alone. It is simply bad design to let one of the competing players unilaterally change the rules of the game. It is no surprise that continental big-C codes, led by Germany after the Second World War, are far more explicit in their attention to parties and their relationship with democracy. Even so, the Anglophone silence is mimicked in comparative constitutional studies scholarship, dominated as it is by American constitutional discourses, including the latter’s attendant pathologies.²⁶ It is almost impossible to properly understand the functioning of different institutional arrangements without close attention to the party system in which they operate.²⁷ Constitutional scholarship that confines itself to institutional analysis alone, without understanding how they are conditioned by political parties, is looking at a seriously distorted picture of constitutional practice.

To some extent, however, big-C constitutional silence on political parties is also explained by the serious conceptual and functional challenges they pose to (liberal democratic) constitutionalism. Conceptually, as the primary holders of real public power in most democracies (at least when in government), political parties should be apt for constitutional regulation. But, as the chief vehicle for organising democratic will, the level of autonomy that political parties enjoy is one of the markers of the health of a democracy. Like private clubs, their membership is voluntary and often informal. The control they have over their members is limited.²⁸ Their role as an intermediary between the state and its people make them insufficiently public to be burdened with the normal duties of state institutions, and inadequately private to be treated on par with rights-bearing citizens.²⁹ Functionally, parties organise popular will and thereby make democratic functioning possible. But as wielders of concentrated state power, they have also emerged as a key threat to liberal democracy (alongside the military and the very rich).³⁰ Constitutions therefore need to walk a fine line between preserving their ability to organise and channel popular will while reducing the threat they pose to democratic governance. Scheppele is right in exhorting constitutional democracies ‘to find ways to support and regulate’ parties, for – she says – ‘the secret to democratic self-preservation may rest in the realization that the party isn’t over yet’.³¹

In this chapter, Section 4.2 will first provide an idealised functional account of political parties and party systems. The idealised account presented in Section 4.2 clarifies what parties do when they function as they should function in a healthy party system of a representative democracy. As such, this is a conceptual and normative, rather than a historical, project. The normative account is acontextual, but the principles offered are pitched at a sufficiently high level of generality to be worth taking into consideration when making all-things-considered decisions in a wide variety of contexts, but not so high and abstract that anything goes. I will argue that parties are difficult to regulate constitutionally because of their Janus-faced public-private character. The key function they perform, when functioning as they ought to function, is to facilitate a mutually responsive relationship between public policy and popular opinion by acting as an intermediary between a state and its people. When they perform this function effectively, political parties significantly reduce four key information and transaction costs that would otherwise make democratic governance impossible: political participation costs, voters’ information costs, policy packaging costs, and ally prediction costs. For critics of this methodological approach who might be worried that an idealised account is too far removed from how parties in fact function in the real world, the practical payoff is that it helps us identify pathological parties and party-systems and diagnose their particular ills. Even as pathological parties abound, especially in recent years, relatively healthy parties have existed and continue to exist around the world. Furthermore, if we started discounting norms simply because they may never be fully realised, our moral universe would become seriously impoverished indeed. I will therefore use this idealised account to ground four principles that can help diagnose the health of a party system and, consequently, principles that constitutions should seek to optimise in relation to political parties and party systems, with a view to avoiding, diagnosing, curing, or mitigating systemic pathologies.

Thus, I argue in Section 4.3 that democratic states (and their constitutions) should respect and optimise four distinct, and sometimes conflicting, political principles in relation to political parties:

1. i. They should guarantee maximum autonomy for the formation, organisation, and operation of political parties, moderated by the restrictions necessitated by their purpose of winning (a share in) state power (for fixed terms) in competitive elections by acting as intermediaries between the state and its people (the ‘purposive autonomy principle’).

2. ii. They should try to optimise the party system such that the total number of serious political parties is large enough to broadly represent every major ‘voter type’, but not so large that the information costs on judicious voters are too high (the ‘party system optimality principle’).

3. iii. They should ensure a separation of parties and the state (the ‘party-state separation principle’).

4. iv. They should discourage the factionalisation of political parties (the ‘anti-faction principle’).

These political principles are drawn from the value of democracy itself. They are likely to bring real world political parties and party systems closer to their idealised form as described in Section 4.2, thereby improving and deepening democratic governance. As such, they should – alongside other relevant political and constitutional norms – inform fundamental constitutional design choices. Retrofitting the regulation of parties through the small-c constitution after key design choices have already been made in the big-C code is, therefore, a mistake. Big-C constitutional silence on parties is as much a regulatory choice as any other and carries significant risks of unintended consequences. In other words, big-C constitutions – as the chief organisational tool for public power in democracies – simply do not have the option of remaining agnostic about the nature and functioning of political parties. The question is not so much whether to regulate parties but why and how.

Another important caveat – especially to the more lawyerly readers – is that not every desirable norm can be converted into law, nor should it be. Some of the implications of the arguments in this chapter are legal, many are matters of political morality. Saying that a norm is a matter of political, rather than legal, morality is not to suggest that there is nothing we can do about its enforceability. Constitutions can make it more likely that certain political norms will be complied with – these are also design choices, they simply transcend the law. In particular, different electoral systems, legislative design, and executive-legislative relations fare differently with respect to the principles defended in Section 4.3. These matters of constitutional architecture have serious implications for the health of a regime’s parties and its party system and are largely settled outside the courts. The aims of this chapter are largely theoretical and conceptual. A more detailed account of what practical implications might follow if constitutions are to take these principles seriously have been canvassed in a different article.³²

4.2 Parties: An (Idealised) Functional Account

In this section, I will argue that political parties, when they function as political parties ought to function, perform the key democratic function of acting as an intermediary between the state and its people in a representative democracy. Two particular features make this intermediary function of parties unique: the bidirectionality of their intermediation and the plenary character of political parties. A party system with healthy functional parties incurs lower levels of four key information and transaction costs: political participation costs, voters’ information costs, policy packaging costs, and ally prediction costs. Keeping these costs low makes a representative democracy viable as a mode of governance. Clear recognition of these features of a healthy party system with functional parties allows us to distinguish them from pathological party systems and diseased parties.

4.2.1 Parties as Intermediaries

A democracy, by definition, requires the rule of (all) the people who constitute a polity. Systems that systemically engage in comprehensive exclusion or suppression of discrete groups of voters, therefore, are diseased democracies, or not democracies at all. If democracy was the only legitimacy criterion for a political regime, we would vest all decision-making powers directly in the people. But democracy is not the only value we care about. Most of us would consider a law requiring the enslavement or genocide of any group thoroughly illegitimate, howsoever democratic its pedigree might be. Less dramatically, rule of law (legality) values – such as fairness, consistency, efficiency, impartiality, non-retroactivity, and generality – invite us to distinguish between rule-making and rule-application.³³ We will put the enduring scholarly debates about the legitimacy, sharpness, and feasibility of the distinction aside for the purposes of this chapter, but note that constitutional practice in liberal democracies has, broadly, come to accept that while the rule-making aspect of political power must be largely vested in institutions that represent the people, rule-application is, on the whole, best left to experts of some description. With regard to rule-making, there are several ways of securing the normative ideal of a popular regime (i.e., a regime that is ‘of the people’ but not necessarily one that is all the rage with the electorate). Direct democracy usually imposes high transaction costs, especially in large and complex societies, although even simple democracies have need for at least some political offices (and committees) that can represent and act in the name of the people and procedures that determine how they may act validly.³⁴ Ancient Athenians used sortition – selection by lot – to fill political offices; in our times, election by universal franchise is the more common method. Political parties have emerged as the main – and arguably indispensable – vehicle for facilitating representative democratic elections in even moderately sized complex modern states.³⁵ What follows is an idealised functional account of political parties in a democracy. Real-world parties will no doubt fall short of these ideals, but the fact that we may never fully achieve an ideal is not a reason to give up on trying to realise them as far as possible.

The chief function of political parties is to act as intermediaries between the state and its people. This claim does not presuppose a specific type of party organisation: I use the term ‘intermediary’ in a loose sense here to be compatible with a varying range of intensity in the relationship between the party and the people.³⁶ What matters is that parties have a threshold level of communicative relationship with the people. State officers and institutions are typically too removed from the people to access popular opinions directly, and ordinary civil society organisations are usually too removed from the state to influence state policies. Exceptions no doubt exist: in systems in which individual legislators represent sufficiently small constituencies, they can have a direct relationship with their constituents; similarly, many policy influencers, such as wealthy corporations, lobbyists, thinktanks, and powerful media houses, often have a significant influence on state policy. Yet political parties are a very special type of intermediary between the state and its people for two reasons: the bidirectionality and the plenary character of their intermediary function.

4.2.1.1 Bidirectionality of Parties

Mediation by parties is bidirectional, in as much as they simultaneously perform both functions of accessing popular opinion and shaping state policy. They are embedded in the structures and institutions of the state but also (at least ideally) have direct access to the people. This simultaneity is essential to the democratic legitimation that parties alone can provide to rule-making state institutions and offices. In general, the state functions through offices and institutions: these are modes of corporate action that are defined by a measure of formalisation of their processes, purposes, and modes of operation. This formalisation is typically necessary for satisfying various virtues (rightly) associated with the state: impartiality, rationality, fairness, legality, and so on. But formalisation imposes a cost – it reduces the ability of offices and institutions to (informally) connect with the people and build authentic interpersonal relationships of mutual understanding and dialogue. Healthy parties, on the other hand, despite their internal institutionalised structures, tend to retain the nimble flexibility and informality of civil society organisations – at least in their local units. This measure of informality allows them to perform their key coordinating function: to imbibe and influence popular opinion, on the one hand, and to formulate and justify their proposed policy package, on the other. The relationship between popular opinions and policy packages is mutually responsive – in a well-functioning democracy, they respond to each other and form a feedback loop. The central task of political parties is to facilitate this responsive relationship between popular opinion and policy.³⁷ Sometimes, they absorb popular opinions and translate them into policy proposals. At other times, they articulate policy proposals and mould public opinion to get behind them.

While Sartori accepts this dual function, ‘grant[ing] that parties are a two-way communication channel’, he insists that ‘the conclusion does not follow that parties are a transmission channel downward to the same extent that they are a transmission belt upward’.³⁸ In part, his reluctance to idealise the bidirectional communicative purpose is based on his characterisation of what I have described as the justification of public policy to the people as ‘manipulation’ of popular opinion.³⁹ This pejorative characterisation under-appreciates the democratic need for the state to justify its policies to the people, including to those who disagree with the government, and the critical role that (healthy) parties play in demanding, articulating, and challenging such justification.

4.2.1.2 Plenary Character of Parties

The second special feature of the mediation role that well-functioning parties play between the state and its people is their plenary character. In heterogenous societies, the values as well as the interests of the people are likely to be diverse. Value pluralism as well as interest pluralism pose a huge challenge to the ability of the state to frame public policy that would be broadly acceptable to its people. The multitude of ways in which different values and interests may combine is so staggeringly large that any complex society faces the potential problem of being left with most of its population being perennially disgruntled. Parties (when they function well) perform a significant legitimation function for the state by coalescing around distinct families of values – often described as ideology – and aggregate the diverse interests of (all) the people of a state into a coherent policy package more-or-less compatible with their ideology. The policy package need not be internally coherent – it often involves the weighing of various interests, preferences, and values. It may entail a multitude of compromises that seek to bridge the gap between the ideal and the feasible and must frequently cater to logically opposed interests, values, and preferences.

The internal contradictions of the policy package of a well-functioning political party notwithstanding, the party can claim that its mediation has a plenary character in three distinct senses: first, it mimics the plenary nature of governance, which is at least potentially concerned with all issues affecting human flourishing (as well as interests of non-human beings). No state can decide to have a policy only on healthcare, for example. Even its silence or inaction on all other matters will amount to a policy decision, which it would be well-advised to adopt deliberately rather than inadvertently. As the drivers of governments-in-waiting, governance parties (that seek to capture high executive offices, as opposed to influence parties that primarily focus on policy impact) come up with policies on a wide range of issues, drawing upon their interaction with the people, and then seek to sell them politically to the people as a package. In doing so, they persuade their supporters to accede to certain compromises made with their own values, interests, and preferences so long as the overall policy package remains attractive to them. These policy platforms also make the opportunity costs of their policy packages transparent to voters, who are better able to prioritise their preferences in the context of resource constraint.

The policy package of a healthy party is also plenary in a second sense: it is one that is designed by putting the interests of all the people on the scales. I will shed further light on this feature when discussing the anti-faction principle. For now, it will suffice to note that a party need not – and cannot – commit itself to all the mutually incompatible values that the multitude of people in a state adhere to, nor can it sincerely claim to have aggregated the preferences of all the people in even a moderately diverse society. But what a non-factional party can – and should – do is to sincerely seek to aggregate the interests of all the people in its polity. No policy package can cater to all these interests equally, and some interests will of necessity be compromised or sacrificed for others. But parties should consider the well-being of all the people: any political group that a priori dismisses the interests of any section of the population as either irrelevant to its policy considerations or worse, meriting its hostility, is no longer committed to the rule of all the people and is basically a faction rather than a party.

Third, parties have a plenary character in as much as they are more likely than most other political actors in electoral democracies to have long-term horizons and therefore are likely to care more about the interests of the future people. Unlike the naturally limited lifespans of individual politicians, parties can – and usually seek to – endure over a long time. Sure, parties care about winning the next election, but that is not their only goal. It is sometimes rational for parties to prioritise ideological victories, organisation-building, performance in future elections, etc. over winning the next election only because parties have longer-term time horizons. As Rosenbluth and Shapiro correctly state, ‘parties have reputations that outlive those of individual politicians, and to the extent that they must represent a wide view of societal interests, they are more capable of delivering desired outcomes than any amount of direct democracy, and more trustworthy than even the most appealing individual politician’.⁴⁰ Indeed, parties frequently outlast constitutions, even several constitutions. This endurance feature adds a temporal dimension to the inclusive plenary character of parties and makes healthy parties much more likely than parties dominated by single individuals to attend to issues such as climate change that will disproportionately affect non-voting children and future people.

4.2.2 Key Costs Reduced by Parties

In providing this uniquely bidirectional and plenary mediation between the state and its people, political parties (in efficient multi-partisan systems)⁴¹ reduce key information and transaction costs for both, making democracy possible. For Sartori, ‘Parties make for a “system” … only when they are parts (in the plural); and a party system is precisely the system of interactions resulting from inter-party competition. That is, the system in question bears on the relatedness of parties to each other, on how each party is a function (in the mathematical sense) of the other parties and reacts, competitively or otherwise, to the other parties.’⁴²

Parties are able to reduce the costs I am about to discuss mainly in well-functioning party systems. One-party ‘systems’, therefore, obviously fail to achieve a reduction in the costs that makes democracy possible. Multi-partisan systems – defined by the nature of the parties they have and the nature of the interaction between them – may be more or less efficient at reducing these costs. Other things being equal, constitutions will deepen democracy if they make their party systems and parties more efficient at reducing the following costs.

4.2.2.1 Political Participation Cost

First, healthy parties in efficient party systems reduce the transaction costs of political participation for citizens (political participation costs). Even in a smallish party-less democracy, an ordinary citizen acting on her own would almost certainly need to take up political engagement as a full-time occupation to have any hope of making a modicum of difference to state policy. Sooner or later, she will have to invent something that looks like a political party to enable some political engagement by citizens who do not wish to become full-time politicians. Parties also reduce the transaction costs of political participation for citizens – not only for partisans, but also for non-partisan citizens – who, in a well-functioning pluralistic democracy, are likely to find some party that reflects their values and priorities most closely and could therefore be their first port of call when raising a matter of political concern.⁴³

High political participation costs can be debilitating for a democracy. Imagine a society without shoemakers. In such a society, one would have to make one’s own shoes or do without them. By reducing the political participation costs, parties offer a similar service of specialisation to citizens – they can custom-build their own mode of political participation, but without parties this is going to be expensive, futile, or both. One of the biggest limitations of recent innovations such as citizens’ assemblies and other party-less sortition-based mechanisms aimed at enhancing political participation is that even as they facilitate political participation,⁴⁴ they are not likely to do so for most citizens for the issues they care about most. A person who is really interested in the question of climate change will not feel sufficiently included without any political party that incorporates the issue on its political agenda, even if this citizen is selected through sortition to participate in an assembly to determine whether abortion should be legalised.

Furthermore, because of their temporally plenary character, parties alone can provide a modicum of representation to future electorates, who are otherwise entirely unable to participate in a democracy. Note that, by definition, the political participation cost has to be affordable by all the people, if the regime is to count as a democracy. Any group that is permanently excluded from the political process because their participation cost is too high changes the very character of the regime. Thus, systems that systemically exclude or suppress voter types, gerrymander constituency boundaries to make their access to power very difficult, or permit political domination by the wealthy through inadequate regulation of campaign finance impose very high political participation costs. This is why, in a healthy party system, parties will be accessible to all persons without discrimination and have a fair chance of winning (subject to the principles explained in Section 4.3); if a salient voter-type finds itself without a political party representing it, the costs of creating a new party to provide such representation will also be relatively low in a healthy system.

4.2.2.2 Voters’ Information Cost

Secondly, parties reduce information costs. In constituencies whose large size is typical of contemporary states, voters tend to lack personal knowledge of electoral candidates. Given modern population sizes and the predominance of gesellschaft relationships outside small kinship and friendship circles, it is usually not feasible to have constituencies so small that most voters are personally sufficiently acquainted with all candidates. Parties reduce the information costs for voters because party affiliations of different candidates provide them with a significant amount of broadly accurate proxy information about their political views and agendas, thereby reducing their voters’ information costs. Party systems that get rid of individual candidates entirely and allow voters to vote directly for parties may be particularly efficient at keeping this cost low, assuming that the number of parties in the system is small enough and their policy platforms transparent enough for a reasonably diligent voter to make informed choices without too much effort. But voter education can be difficult in closed list voting systems because it is not just parties that supply proxy information for their candidates; local candidates also educate voters about their parties. A system where both parties and local candidates (in single-member or multi-member constituencies) matter is likely to reduce voters’ information costs to the greatest extent.

It is easy to take the lowering of voters’ information cost by parties for granted, but in doing so we imperil democracy itself. When electing a government, the act of voting is not merely self-regarding but also other regarding. Given the serious implications my vote can have on the lives of others, it is reasonable to suppose that the right to vote also entails a (moral) duty to vote judiciously. Casting a vote is more like a judge deciding a case (albeit without the duty to give public reasons) than choosing what to eat from a restaurant menu. While a voter has a duty at least to discharge her democratic function with due diligence (if not also with an attitude of care towards her compatriots),⁴⁵ it is also incumbent upon the state to ensure that the diligence burden on a citizen trying to vote judiciously is not too onerous. One implication of this is that the voters’ information costs should be kept relatively low.

4.2.2.3 Policy Packaging Cost

Third, parties also reduce information costs for democratic state institutions by revealing to them what combination of policies will be acceptable to what proportion of the people. All parties that campaign on policy packages provide this information to state institutions, whether they win or lose. And winning parties, in addition, inform state institutions about the particular policy packaging that a large proportion of – if not a majority of – the people are willing to at least tolerate. This information can be generated and revealed, and state policy be legitimised, only through the bidirectional and plenary character of the mediating function that parties perform. Let us label these information and transaction costs as policy packaging costs.

The importance of lowering the policy packaging costs for state institutions should be obvious. It allows them to be responsive to the people in a way that is likely to cause the least amount of disgruntlement among those who disagree with individual policy proposals. Bundling different proposals into a single policy platform, while also rejecting alternative policies, makes the opportunity costs of policies clear to voters. They know not only what they are getting but also what they cannot get if they choose a certain set of policies.⁴⁶ This is key to the legitimation of the state’s policies to its people. The fact that the policy packages are themselves framed bidirectionally by political parties drawing upon their proximity to popular opinion, and then justified to the people, buttresses the perceived as well as actual democratic legitimacy of partisan state institutions. Without parties, there is simply no efficient way for state institutions to gather this information: opinion polls can tell us the level of popular support for particular issues, but the unity that packaging brings to a stack of proposals is only possible through the iterative platform and the legitimising glue that only a political party can supply. In other words, in a party-less state, the regime-legitimising policy packaging costs are infinite.

4.2.2.4 Ally Prediction Cost

Finally, parties reduce information costs for other political parties as well as for state officers and institutions by indicating to them which office-holders are likely to be persuadable political allies, whose support can be taken for granted, and who are likely to oppose certain policy proposals. Moreover, when they are reasonably disciplined, parties permit the identification of key leaders whose support will likely translate into the support of a predictable number of legislators and what it might take to secure their support. By aggregating and publicising political leanings, parties reduce the information costs associated with discovering whether another political actor is a political friend or foe and the consequent transaction costs in making political decisions (ally prediction costs).

Ally prediction costs are also of considerable significance in a democracy. Without parties, any decision-maker will need to seek the individual consent of every representative. Because policy packaging will also be absent, the ensuing political bargains will need to be simultaneous, numerous, and mutually compatible. Every individual representative in such a system will have the incentive to maximise the concessions she can extract for her support, whereas the policy initiator must make these concessions until she has the necessary support – she will, after all, be working with little knowledge of potential allies given the high ally prediction costs. The number of possible veto-players will be too high, and thus, state policy will tend towards preserving the status quo. Democratic decision-making under such scenarios can become extremely difficult.

We can therefore see the important role parties can play in keeping these four key information and transaction costs low. Without them, these costs will be too high to permit the smooth functioning of a democracy . Recent proposals for directly democratic citizen’s assemblies (especially when conceived as decision-making rather than advisory bodies) grossly underestimate – or ignore – the substantial hurdle that these costs pose and the role that political parties play in reducing them and making them affordable.⁴⁷ Parties are therefore the key vehicle for a responsive interaction between public opinion and public policy, the very essence of democratic governance. Apart from policy responsiveness, the intermediary role that parties play makes them ideal training grounds for tomorrow’s leaders, for public education and debate on civics, and as a vehicle for social solidarity and camaraderie. The training role is often underestimated in our popular culture that does not appreciate that politics is a skilled profession, like every other, one that typically requires a politician to regularly meet political opponents who vehemently oppose and criticise them, to interact with a wide cross-section of society that includes people from backgrounds very different from their own, to be nimble with compromises to get things done, to be able to withstand the emotional burden of constant public scrutiny, to learn to speak pithily and engagingly, to respond quickly to changing circumstances. These virtues do not come easily to most of us; parties provide a forum for acquiring and honing these key political skills and a community for solidaristic support and mentoring from party colleagues that these difficult engagements often necessitate.

With respect to citizens, the acts of raising a concern with the local party representative, becoming a party member, campaigning in an election, or joining a political protest organised by her party transform a legal subject into a performative citizen. Healthy parties can translate an individual’s grievances into common causes, allowing her to see, for example, that the absence of a girls’ toilet in her daughter’s school is a broader concern that implicates patriarchy and affects women and girls across the country.

In this section, we have seen that political parties are essential to the proper functioning of representative democracies in sufficiently large and complex polities. They act as intermediaries between the state and its people, on the one hand, transmitting popular opinion to state institutions that typically lack the ability to gauge it directly and, on the other hand, formulating state policies and justifying them to the people. This dual role gives them a Janus-faced public-private character – they need to operate as a private association proximate to the people in order to access popular opinions and justify state policies. They also simultaneously need to be embedded in (but not fused with) the institutional structures of the state to transmit popular opinions back to them and to help them formulate policies, which in turn they will help justify to the people.

Healthy parties in well-functioning party systems therefore grease the wheels of representative democracy by reducing the following information and transaction costs: the political participation costs and the voters’ information costs for the people and the policy packaging costs and ally prediction costs for state institutions. This functional account of political parties has highlighted their indispensability to the effective functioning of a representative democracy. This alone should suffice to alert us that lazy dismissals such as ‘all politicians are corrupt’ and ‘all parties play dirty’ imperils democracy itself. While a discursive defence of the importance of parties in a democracy is crucial, the state and its constitution too can lend a helping hand. They can bolster the ability of individual parties to perform their important mediating function between the state and its people as well as structure the party system in ways that facilitate rather than hinder the ability of parties to deepen democracy. They are more likely to do this if they deliberately consider certain guiding principles in constitutional design.

4.3 Constitutional Principles in Relation to Political Parties

Section 4.2 offered an idealised account of what parties do in a well-functioning democratic system. This idealised account is helpful in distinguishing between parties and party systems that function reasonably well from those that are pathological. It also helps us diagnose the particular ailments that afflict a party or a party-system. It should be clear that parties that fail to perform their intermediary function appropriately and effectively are bad for democracy. A party system can tolerate a few malfunctioning parties and still be healthy, so long as most of the governance parties are sound. In this section, we will turn our attention to certain normative goals that constitutions ought to adopt in relation to parties that will make it more likely that parties and party systems are healthy or that – if there are diseased parties in the system – the system can still tolerate or mitigate their ill effects or even nurse them to better health. To do so, I will explain four principles that constitutions should adopt in relation to parties: the purposive autonomy principle, the party system optimality principle, the party-state separation principle, and the anti-faction principle. Readers should note that for reasons of space, this already ambitious chapter does not take the logical next step of outlining more concrete prescriptions that might follow from the adoption of these principles. Some of that discussion has been undertaken elsewhere.⁴⁸ In this section, I will argue that democratic states should, through their constitutions, consider and seek to optimise four distinct, and sometimes conflicting, political principles in relation to political parties:

1. i. They should guarantee maximum autonomy for the formation, organisation, and operation of political parties, moderated by the restrictions necessitated by their purpose of winning (a share in) state power (for fixed terms) in competitive elections by acting as intermediaries between the state and its people (the ‘purposive autonomy principle’).

2. ii. They should try to optimise the party system such that the total number of serious political parties is large enough to broadly represent every major ‘voter type’, but not so large that the information costs on judicious voters are too high (the ‘party system optimality principle’).

3. iii. They should ensure a separation of parties and the state (the ‘party-state separation principle’).

4. iv. They should discourage the factionalisation of political parties (the ‘anti-faction principle’).

As constitutional principles, they primarily address the state. However, because the ruling party/coalition, which typically holds the reins to state power, is unlikely to be particularly interested in respecting them – and may even have a strong self-interest in breaching some of them – these principles need to be constitutionalised as pre-commitments by the state. I hasten to add two caveats to this proposal: first, I do not take constitutionalisation to necessarily entail judicialisation.⁴⁹ In fact, sometimes it may be neither necessary nor desirable to express a constitutional principle as a constitutional norm directly regulating constitutional actors, let alone as a legal norm. Instead, establishing an institutional arrangement that is most likely to uphold that principle – what may be termed ‘second order’ regulation – may well be the most optimal design solution.⁵⁰

Second, a norm can be ‘constitutionalised’ in multiple ways, its inclusion in a big-C constitutional code being only one of them. Other modes of constitutionalisation include judicial interpretation, quasi-constitutional statutes, and constitutional conventions. The proposed principles should ideally be reflected – at least at a broad level – in the big-C constitutional code so that the institutional arrangements of the state are framed alongside its party system rather than ex ante. The finer details will, obviously, need to be left to the small-c statutes, conventions, and caselaw. The key determinant in each context should, in the main, be feasibility and effectiveness in light of path dependencies and all-things-considered judgments. The following sub-sections will explain each of these principles in turn.

4.3.1 The Purposive Autonomy Principle

4.3.1.1 The Public Purpose of Private Parties

States and their constitutions must seek to support (existing and future) political parties in the performance of their bidirectional and plenary intermediary role between the state and its people, so that they are best placed to reduce the four key information and transaction costs that make democratic governance difficult. In order to do so, parties must remain simultaneously private as well as public – this duality is their unique strength and also the reason for the complexity in regulating them. Liberal constitutionalism has long adhered to a controversial public-private divide. This divide is premised on diametrically opposed default assumptions about the regulation of public and private actors. Private actors are granted the autonomy to do whatever they like, unless there are very good, and special, reasons for regulating their actions. Public actors, on the other hand, may not do anything at all unless they have good, and constitutionally permitted, reasons for doing something. This distinction has long been criticised, with some scholars calling for its abolition,⁵¹ whereas others have argued that the distinction may be preserved while the line between the public and the private should be drawn differently.⁵² In a previous work, I have argued that the distinction is best understood as a spectrum rather than two discrete boxes.⁵³ The spectrum is both actor-sensitive and action-sensitive and primarily tracks interpersonal power and other-regarding functions. At the public end of the spectrum stands the all-powerful state enacting a general criminal or tax statute in its legislature. At the other end, a natural individual person lost in deep thought in her bedroom is paradigmatically private. When she acts as the manager of a large firm the next morning, she has moved away from this private-most end and become more public. Some areas of law – such as discrimination law – have come to terms with the idea that the public-private divide is a spectrum.⁵⁴

Constitutional law is yet to follow this trend. It continues, on the whole, to draw the line sharply, vesting private actors with constitutional rights and burdening public actors with constitutional duties.⁵⁵ This structural limitation is an important hurdle that must be overcome if constitutions are to properly regulate political parties without destroying their public-private duality. Treating them as just another state institution is likely to seriously compromise their ability to engage with the people directly. While constitutions must be careful about over-regulating political parties lest they destroy their private character, they should also worry about constitutional silences and under-regulation that fails to acknowledge their publicness. A fit-for-purpose constitutional scheme for political parties will pay attention to three dimensions: (i) subject to the principles discussed in this chapter, it will grant them maximum autonomy; (ii) it will vest in them the necessary rights, powers, and entitlements that will enable them to better discharge their functions; and (iii) it will impose only those duties on parties that are necessary to preserve their public character. Is there such a happy regulatory middle that would preserve their privateness while demanding that they be sufficiently public at the same time?

To locate that regulatory middle, we need to point out with greater precision what precisely makes parties public. The private dimension of parties demands maximum autonomy for the formation and operation of political parties. But their public character demands a recognition of their purposive dimension: unlike natural individuals, political parties in a representative democracy cannot be allowed to choose their purpose with complete freedom. What makes them a political party in a democratic party system is their public purpose of participating in competitive elections – with other parties – in order to secure a measure of control of the levers of state power for fixed periods of time and to do so by acting as intermediaries between the state and the people. This purpose is definitional of what a political party in a democracy is. It is specified at a high level of generality, being compatible with an extremely wide range of more specific purposes that parties may have. But it is incompatible with certain purposes: such as instituting a single party state; making elections insufficiently competitive; barring or making it difficult for (other) parties to connect or communicate with the people; and so on.

4.3.1.2 The Autonomy of Parties

To respect their private dimension, constitutions should guarantee maximum autonomy for the formation (from scratch or by splitting an existing party), organisation, and operation of political parties, moderated by the restrictions necessitated by their public purpose of winning (a share in) state power (for fixed terms) in competitive elections by acting as intermediaries between the state and its people. Hence the purposive autonomy principle. Simply put, the principle permits significant autonomy to parties (and partisans) but seeks to ensure that they are committed to the purpose of being but one player in a multi-party democracy. The principle requires that parties should be relatively easy to form and disband and to enter or leave. The main barriers to their success should be political, not legal. New parties or opposition parties must not be locked out of political competition through high entry barriers.⁵⁶ High access barriers are not only bad for new parties but also for established parties – without an alternative political outlet, strong political forces that are excluded from partisan expression would eventually seek to capture an established party or upend the system itself. The autonomy of political parties is key to keeping the political participation costs in a democracy low.

Translating the need to protect their purposive autonomy into particular rights and duties requires further specification beyond the scope of this chapter. Full specification cannot, obviously, happen without catering to the peculiarities of a particular political context. In general, parties may need the whole suite of civil and political rights that citizens ordinarily have access to in a liberal democracy; sometimes they may even need special protections of their autonomy over and above what citizens are guaranteed. They are likely to need access to all the usual fundamental freedoms to enable them to be formed, to contact and connect with the people and mobilise them, to campaign and express political views, to contest elections, to raise funds, and so on. Without these freedoms, a political party may be woefully inept at reducing key democratic costs. Furthermore, a range of ‘fourth branch’ guarantor institutions that can effectively guarantee norms such as free and fair elections, probity, transparency, fair boundaries delimitation, campaign finance regulation, and a host of other norms that the ruling party/alliance may have reasons to want to undermine are usually necessitated by the purposive autonomy principle.⁵⁷ These constitutional protections of party autonomy are also necessary to insulate opposition parties from any self-interested targeting by the ruling party/coalition: not least because the autonomy of all (serious) parties matters, such that the principle of inter-party equity is embedded in the purposive autonomy principle.

4.3.1.3 Public Entitlements for Serious Parties

While their privateness demands the protection of their autonomy, their public purpose may entitle them to special privileges and powers, as well as make them fit for bearing special duties that are inapplicable to natural individuals. Public entitlements, such as (limited) state funding for political campaigns or immunity from defamation laws for political speeches, can help secure a level playing field between political parties and enable many of them to discharge their democratic functions effectively.⁵⁸ This would be especially useful for smaller parties in a hegemonic or a predominant party system.⁵⁹ Even if the predominant party is itself healthy, the party system isn’t. Thus, any provision of state benefits has to avoid cartelisation. In order to reduce the political participation costs of all voters, a measure of state support to less powerful parties would usually be necessary. That said, it may be permissible to restrict these rewards to serious political parties. Seriousness is a measure of the party’s intention to play the requisite intermediary role between the state and its people – a serious political party seeks a role in the governance of the state, or an influence in state policies, or both, primarily by winning elections and shaping political discourse. Note that both governance parties and influence parties are ‘serious’ in the sense I intend here. On the other hand, a party that doesn’t campaign or put up candidates for elections hardly has a public dimension worth worrying about. Similarly, joke parties (such as the British Monster Raving Loony Party) and a number of single-issue parties (such as the Australian Help End Marijuana Prohibition Party) often don’t even intend to win – and tend to be themselves surprised when they do win – even if their ability to focus the spotlight on an ignored issue can often be valuable for a polity. Such non-serious parties can largely be regulated as private clubs left to enjoy the autonomy that other political parties enjoy, but without the public benefits or burdens until they have proven their political seriousness.

While performance in past elections is a reasonable proxy of seriousness, it is generally a good idea to base it on aggregate performance in the three or four previous elections taken together, rather than the most recent election alone. For example, equitable state funding may be provided to every party that secured at least (say) 5 per cent of the popular vote share in any of the last three election cycles, or to every party that had a presence in the legislature in at least one of the last three elections. Where these seriousness thresholds are fixed is necessarily arbitrary and context-specific – any reasonably low threshold that has not been weaponised against particular parties and doesn’t squeeze out smaller, but serious, parties should normally suffice (any such weaponisation would, we will see, breach the party-state separation principle). Admittedly, one problem with measuring seriousness (a forward-looking phenomenon) through past electoral performance is that it cannot distinguish between serious and non-serious new parties. Since the purposive autonomy principle applies not only to existing parties but also to future ones, any forward-looking public support may need to be extended to all new parties willing to accept the public duties imposed on serious parties, with exclusions kicking in after they have participated in (say) two election cycles.⁶⁰ Note also that while seriousness of a political party is a relevant consideration in determining public funding, it does not extend to calibrating the extent of funding proportionate to party size. If anything, the kind of distinction US law draws between major and minor parties in order to supply greater benefits to the major parties are likely to be inimical to the principle of equity embedded in the purposive autonomy principle (as well as the party-state separation principle).⁶¹

4.3.1.4 Public Duties of Parties

The public purpose of parties invites not only special entitlements but also some public duties. We will discuss some of them under the other principles to follow. But some duties flow directly from their role in reducing the political participation costs and the policy packaging costs. Parties are likely to lower the political participation costs only if there is a fair measure of transparency surrounding their core value commitments, internal institutional structures, decision-making processes, financial affairs, and credible – even if internal – enforcement mechanisms of their institutional commitments. Serious parties should, therefore, have the duty to publicise these details, so that voters and partisans can rely on them and hold parties accountable. Likewise, parties are likely to lower the policy packaging costs (as well as the other three democratic costs) only if they offer a more-or-less comprehensive policy package in their election manifestos. Serious parties may, therefore, be mandated to adopt and publicise (at least broad) policy positions on a number of the key issues of governance of the day (such as taxation, foreign policy, education, health care, immigration, and so on) in their election manifestos, preferably published before a fixed period prior to elections to give them sufficient public airing.

On the whole, the purposive autonomy principle seeks to preserve the public-private duality of political parties that is essential to their role in facilitating democratic governance. Supporting parties requires guaranteeing considerable autonomy to all existing and future parties. At the same time, at least serious political parties need affirmative state support and protections that enhance their abilities to reduce the four democratic costs. The purposive autonomy principle also justifies the imposition of the duty to adopt transparent party constitutions and plenary policy manifestos on such serious parties. But all public duties imposed on parties need to be justified with reference to their public purpose of being but one player in a multi-party democracy. If we consider Article 21(1) of the German Basic Law, for example, it is broadly a recognition of the purposive autonomy principle:

The political parties participate in the formation of the political will of the people. They may be freely established. Their internal organisation must conform to democratic principles. They must publicly account for their assets and of the sources and use of their funds as well as assets.⁶²

‘Broadly’, because I am doubtful that inner-party democracy – mandated by the third clause above – can be justified by the purposive autonomy principle. It is by no means obvious that internally democratic parties are better at reducing the key democratic costs, not to mention the pragmatic difficulties in determining what suffices as an internally democratic party at an age of relatively loose and myriad ways of associating with a party.⁶³ If anything, base-voter-dominated party primaries have emerged as a key centrifugal force in US politics and the concomitant threat that force poses to democracy in the country.⁶⁴ Constitutions should be slow to mandate inner-party democracy or regulate how parties discipline their members. Many courts have enforced fundamental rights claims by ordinary voters and party members against political parties and their leadership.⁶⁵ Doing so has clear, and often adverse and unintended, consequences for the purposive autonomy of political parties. The implication is not that parties should be allowed to treat their members in any manner they wish; it is rather that state regulation of the relationship between parties and their members should cross a high bar of purposive justification that takes their hybrid public-private character seriously.

4.3.1.5 Preference for Second-Order Regulation

The need to protect purposive autonomy of parties dictates not only a cautious approach to imposing duties on parties but also how any duties may be imposed. Duty-imposing norms should be crafted so as to not destroy the dual character of parties. In general, and subject to their effectiveness in a given context, three broad regulatory criteria should govern design possibilities for duty-imposing norms with respect to political parties:

• – Political enforcement and self-regulation are better than judicial enforcement,⁶⁶

• – Nudges are better than command-and-control,⁶⁷ and

• – Carrots are better than sticks.⁶⁸

These criteria are partial to ‘second-order regulation’, which emphasise the importance of ‘background competitive structures’ that shape decision-making, rather than seeking to police behaviour directly through first-order commands.⁶⁹ Note that all background structures shape the behaviour of actors – the question is not so much whether to have second-order regulation but what type of second-order regulation is worth having. For example, a democracy has to choose some electoral system, and each system shapes the behaviour of politicians differently. In fact, the choice of the electoral system (majoritarian or proportionate, ranked or unranked, at large or constituency-based), the nature of executive-legislature relations (parliamentary, presidential, semi-presidential, semi-parliamentary), the number of legislative chambers, and the degree of centralisation or federation are all institutional choices that can significantly impact the nature of the party system in a polity. For example, majoritarian, cumulative, approval, and ranked voting systems are likely to incentivise centripetal parties, whereas proportionate representation is more conducive to factions. Parliamentary systems are likely to encourage collective party leadership, whereas presidential systems could encourage individual-centric parties. Bicameralism and federalism are probably better suited to protecting opposition rights than unicameralism and centralisation. The directionality of these precise connections is beside the point – if one concedes that a significant connection exists, in whatever direction, then many regulatory objectives in relation to parties can be achieved by the right combination of institutional design of state bodies. This indirect, second-order, regulation is generally more conducive to party autonomy than first-order legal regulation.

I am not suggesting that rule-based, judicially enforced, command-and-control first-order regulation will never be appropriate. Sometimes it will be. The party-state separation principle will normally require stringent, often legal, norms to protect the separation of the state from the ruling party/coalition. Most aspects of private law, such as the law of torts or contracts, should apply to parties, unless their public character demands an exception to be made. Anti-discrimination norms that support the anti-faction principle are usually already calibrated to justify judicial intervention against hybrid public-private actors⁷⁰ and will therefore be appropriate for judicial enforcement against parties in most jurisdictions. Consider Smith v Allwright, where the United States Supreme Court first determined that all-white primaries for the selection of Democratic Party candidates were so well-integrated in the electoral system of Texas that they amounted to state action, before applying the anti-discrimination principle to the Party and prohibiting all-white primaries.⁷¹ This move may have been doctrinally necessary because of the quirks of the American legal system, but morally speaking, there is no reason why anti-discrimination norms should not apply to parties qua parties, that is, as hybrid public-private bodies, without needing to characterise them as state institutions. In general, however, where feasible and effective, second-order regulation is more likely to preserve the dual character of political parties. Extreme measures such as party bans are best avoided. Courts understandably struggle to condemn even the most clearly anti-democratic parties when the outcome is as far-reaching as a total ban.⁷² What’s worse, even when successfully deployed, party bans may be ineffective, or even counter-productive.⁷³ Having said that, second-order regulations would sometimes fail to achieve their desired purpose. While an independent guarantor institution like a boundaries commission may well ensure fair constituency determination in most cases,⁷⁴ even the best design can fail. If a captured boundaries commission produces a gerrymandered electoral map, there may well be strong reasons for first-order judicial intervention.⁷⁵

The purposive autonomy principle is a meta-principle that dictates how constitutions should approach political party regulation. It demands a move away from a binary dichotomy that liberal constitutionalism is used to: rights for private persons, duties on public bodies. Instead, it draws attention to the need for a more nuanced approach, one that considers each right and duty in terms of its appropriateness for the hybrid character of political parties. The discussion above is not meant to be too directive: in constitutional practice, context matters a lot. Even in a given context, there may be many different ways of satisfying the purposive autonomy principle. The main point is that instead of trying to fit political parties into a ready-made template designed either for private individuals or for state bodies, an a la carte regulatory framework that works for them is required. The three following principles may be understood as facets of the purposive autonomy principle but merit separate discussion because of the important bearing they have on the constitutional regulation of parties.

4.3.2 The Party System Optimality Principle

Healthy parties tend to attract members, affiliates, and voters through their ideologies, rather than through clientelism or patronage. I understand ‘ideology’ in the sense that Converse explains it: a relatively wide-ranging belief system, which is relevant to political behaviour.⁷⁶ A belief system, in turn, is ‘a configuration of ideas and attitudes in which the elements are bound together by some form of constraint or functional interdependence’.⁷⁷ The centrality of an element in a belief system is a measure of the likelihood that a voter will change her party preference if her party’s stance regarding that element changes, rather than change her view on the element itself.⁷⁸ She may tolerate a party’s change of position on less central elements in a belief system but give up on her partisan loyalty if the party reneges on a more central element. For Gerring, the quality of being ‘bound together’ (which he calls ‘coherence’) has two corollaries: contrast (‘implying coherence vis-à-vis competing ideologies’) and stability (‘implying coherence through time’).⁷⁹ Thus, competing political ideologies straddle the same ideological axis and are relatively stable over time. While much penumbral content of political ideologies is malleable, their most central elements are likely to be most relevant to contrasting them with other ideologies and determining their stability over time. Not all ideologies matter politically. Chhibber and Verma argue that politically salient ideologies not only need to be competing and stable but also possess two further features to structure a party system:

First, there must be political, social, or economic elites with interests, vested or otherwise, who differentiate themselves on the basis of such ideas, offer resources to support the creation of a particular ideological position, and assist in the transmission of these ideas to the voters. And, second, the ideas thus transmitted, and the issues that embody them, must have the support of enough people.⁸⁰

Without sufficient elite support, parties are unlikely to get off the ground. Without enough support in the electorate, they are unlikely to become meaningful political players. Sartori’s classical account analysed party systems through this lens of salient ideological axes. His distinction between two-party systems, moderate pluralism systems, and polarised pluralism systems was based on two factors: the number of political parties in a system and the ideological distance between them.⁸¹ In this model, a two-party system had two large, ideologically centrist parties (i.e., very little contrast between their ideologies); a moderate pluralism system had multiple parties organised around two broadly centrist ideological coalitions; and a polarised pluralism system had multiple parties pursuing ideologies cutting across the traditional left-right axis seen in the other two systems. The ideological distance between the parties/coalitions in the two-party system and the moderate pluralism system was relatively modest because each pursued the median voter organised on a single – macroeconomic policy – axis. On this account, the first two centripetal systems usually yielded stable and effective governments, whereas polarised pluralism often resulted in political as well as democratic instability.

This classical account needs an important revision in our times. As Scheppele has argued, politics is no longer organised on a single left-right ideological axis in contemporary Western democracies. In the very least, a nativism-cosmopolitanism divide has strongly emerged as an additional, cross-cutting, axis for political alignment.⁸² Chhibber and Verma argue that independent India has always had two salient political axes: the politics of statism (‘the extent to which the state should dominate society, regulate social norms, and redistribute private property’) and the politics of accommodation (‘whether and how the state should accommodate the needs of various marginalised groups and protect minority rights from assertive majoritarian tendencies’).⁸³ These insights scramble the tidiness of Sartori’s single-axis classification. When two major ideological axes are salient to voters, there are at least four stable party types (and concomitant ‘voter types’) that can broadly capture the worldviews and political preferences of most voters in such systems: on Scheppele’s classification, for example, one should expect left-nativist parties,⁸⁴ right-nativist parties,⁸⁵ left-cosmopolitan parties,⁸⁶ and right-cosmopolitan parties.⁸⁷ With each new salient axis, new permutations give rise to the possibility of an even larger number of voter types in search of distinctive political representation.

These ideological axes must be salient: political parties offer package deals, and they cannot be expected to customise their policy offerings to all the individual preferences of every voter. We should therefore understand key voter types only in relation to the politically salient ideological axes of division in a given polity, which in turn are defined by the most central belief elements in their belief system (such as nativism or redistribution). A salient political division comes to define the political preferences of voters in such a deep way that it becomes difficult for the same party to simultaneously represent the preferences of distinct voter types effectively. Such parties may try to speak in multiple voices to mutually incompatible constituencies (in the short term). However, if the salient divisions are abiding, these parties are either likely to split (if the system tolerates multiple parties) or have one group eventually come to dominate or decimate the other within the party (if it doesn’t).

If the latter happens, and no party exists to cater to a particular voter type, the political participation costs of such voters will be extremely high, to the point that they may be totally excluded from representative politics. What is worse, even their efforts to start a party that caters to their voter type may be frustrated if the party system is structurally predisposed to a two-party system and does not facilitate the emergence of new parties. Their only remaining options will be either to capture one of the existing parties (and thereby deprive a different voter type of political representation) or try to change the party system, or even the political system entirely, from the outside. Such excluded voters become especially vulnerable to a quick-fix populist rhetoric.⁸⁸ Such powerful, but unspent, political force is extremely dangerous to democracies. Furthermore, the near impossibility of political alliances in a two-party democracy – except when party discipline is extremely weak – is also not necessarily a good thing for democracy. This is the reason why a two-party system is more likely to encourage compromise-resistant tribalism in politics, accentuating the winner-takes-all feature of majoritarian politics.

What follows is that in any system that has more than one salient political axis, a two-party system simply cannot approximate the broad political worldviews of major voter types. A regime will establish ‘the rule of the people’ only if it facilitates the representation of the preferences of every major voter type in its party system, with two caveats: first, as I will argue later in this chapter, it is legitimate – albeit sometimes unwise – to restrict the likelihood of political representation – or, at least, the likelihood of political success – of factional voters who do not accept that a democracy is the rule of all the people, even if a factional-inclusivist axis has become salient in that polity. Just as I cannot rely on my autonomy to sell my children or my (future) self into slavery, rule of all the people cannot be relied upon to transform a democracy into the rule of some of the people. Neither autonomy nor democracy apply to themselves in this self-harming manner.⁸⁹

The second caveat is that there is a feasibility limit to the total number of serious parties that a democracy can accommodate. It is true that the larger the number of distinctive parties in a system, the smaller the political participation costs are likely to be for a voter. In fact, if there is a party that mirrors every voter’s customised set of political preferences, political participation costs will be non-existent for every voter. Needless to say, such single member ‘parties’ won’t be parties in any meaningful sense. Furthermore, even as they reduce political participation costs, a large number of parties significantly increases voters’ information costs. A voter who has to go through a list of fifty candidates belonging to fifty different serious parties is able to make an informed choice only after putting in considerable effort to educate herself on the distinctive ideological commitments and political platforms of all these fifty parties. She might as well focus her research on the fifty individual candidates in such cases (which would not be any less daunting, in any case). Too many choices may not matter when the stakes are low – such as when one is ordering a meal from a restaurant’s menu – for one can make a reasonable choice having considered only the first five options. But when the stakes are as high as entrusting the government of one’s polity, the voters’ information costs must be reasonable enough to enable a judicious voter to consider the pros and cons of all candidates.

The sum of these concerns is the ‘party system optimality principle’: in contemporary democratic polities that divide along multiple salient axes, party systems should be optimised such that the total number of serious political parties is large enough to broadly represent every major ‘voter type’, but not so large that the information costs on judicious voters are too high. The choice of electoral system for the elected institutions and offices is probably the single most important regulatory tool to respect the party system optimality principle (although the party system can also become federalised in federal systems, since province-specific parties can emerge irrespective of the electoral system). Note that a polity can coherently adopt different electoral systems for different state institutions (especially in bicameral and federal systems), thus vastly expanding the regulatory permutations and combinations at its disposal.

4.3.3 The Party-State Separation Principle

One danger in approximating the democratic ideal through elections is the abiding possibility that we only manage to secure the rule of some of the people at any given point in time. The political exclusion of the losing parties – and potentially their voters – is more acute in winner-takes-all systems. We seek to solve this problem by temporally distinguishing between a state’s regime and its government: a regime can still call itself democratic if, over time, it allows different parties to win political power. So long as there are no permanent winners, the regime can still claim to be ruled by all the people. This is the ideal that legitimises many power-sharing arrangements in deeply divided societies.⁹⁰ It also generates our third political principle: that a state should seek to ensure a separation of the ruling party/coalition and the state, so as to allow a genuine hope for today’s losers to be tomorrow’s winners. We will call this the ‘party-state separation principle’. The basic argument is that if a party (usually the ruling party/coalition) becomes entrenched in the apparatus of the state, the political participation costs of the supporters of all other parties become insurmountable.

The party-state separation principle demands a recognition of a host of opposition rights: including a significant opposition voice – perhaps even a veto – in constitutional amendments and constitutional appointments. It requires the bureaucracy, police, prosecution, judiciary, and guarantor institutions to function in a non-partisan manner. The principle also demands equity in state benefits given to the ruling party/coalition and to opposition parties – making it structurally difficult for a given party or parties to win state offices breaches the party-state separation principle as much as making it more likely for a given party or parties to win or retain state power. Recognising this principle is especially important given the salience of the institutional ‘separation of powers’ principle in constitutional theory – given how partisan loyalties can scramble institutional separation, it is essential that the party-state separation principle is considered alongside the institutional separation of powers principle and given the same weight in constitutional thought. Note, however, that the party-state separation principle does not require that the opposition be given the right to veto ordinary (as opposed to constitutional) policy objectives of a legitimate and democratically elected government. A balance has to be struck between allowing the opposition to perform its constitutional duties without thwarting the government’s ability to govern.⁹¹

One way to appreciate the party-state separation principle is to think of it as a constraint imposed on the democratic rights of today’s people by those of a future people. Constitutional theory has frequently worried about ‘the dead hand of the past’ in the context of entrenched fundamental rights but rarely considered constraints generated by the future people’s right to democracy. Even if we assume that there are a people at a given point in time in a state, and that this people today unanimously wish to entrench the ruling party/coalition in the apparatus of the state, they cannot possibly obtain the consent of any future peoples to do so and must therefore lack this power. One might seek to respond to this claim by suggesting that it might at least be permissible to fuse a state and its ruling party/coalition for one generation, and the arrangement may be revisited by every future generation? The claim presupposes the all-too common, but mistaken, generational outlook to think of a ‘people’ in the temporal sense. A people – in its temporal sense – is not like The Doctor (from the famous British science-fiction show Doctor Who), who disintegrates and regenerates into a new version of him/herself at precise moments in time. Each version is the same eternal Doctor, but different, with clear and distinct temporal geneses and dissolutions. A people would be like The Doctor if everyone in a generation was born on the same day and died on the same day. In reality, a people – temporally – are better compared to a river. An ancient Greek philosopher, Heraclitus, is said to have posed the famous paradox: Can one step in the same river twice?⁹² The paradox lies in the fact that anyone stepping into any given point in the course of a river twice – the two attempts separated by some interval of time – will be stepping into different waters of the same river. Just like a constantly flowing river that receives new waters from its source and loses old waters to the sea, a people are constantly gaining new individuals by birth or immigration and losing old ones to death and emigration. Even the separation of a day, therefore, will suffice to separate the current people from a future one. As an ideational entity, they remain the same people over time; as a sum of their constituent individuals, they are constantly changing over time. Thus, even a single generational fusion of the ruling party/coalition and the state will be impermissible by the democratic ideal.

The party-state separation principle therefore requires that a state should preserve the genuine likelihood of different parties securing governmental power at different points in time. The transfer of power following elections should be peaceful, and the political opposition must be able to plausibly imagine itself as a government in waiting. It should therefore be hostile to a one-party system (where only one party is allowed to exist, de jure and de facto), or a hegemonic party system (where smaller parties are allowed to exist, but the system de facto and de jure favours a hegemonic party that remains in power) at all times.⁹³ It should even be hostile to the kind of two-party system in which the two parties operate like a cartel and make it structurally difficult for a third party to emerge.⁹⁴ Any such fusion of parties and the state is not only bad for democracy, it is also likely to make the regime unstable because any significant voter type without mainstream political representation is likely to find solace in anti-system parties. On the other hand, it is compatible with a predominant party system, where a single party or coalition de facto dominates all others, although de jure the system permits free and fair political competition and gives no structural advantage to the predominant party. That said, the purposive autonomy principle would still view a predominant party system as non-ideal and seek to enable opposition parties to rise and flourish in such a system. Even if the opposition does not win elections, a robust opposition is essential to check the political power of the ruling party/coalition (balanced against the need for effective government) and therefore to reduce the four democratic costs effectively. Recall that these costs remain high in a system with only one healthy political party. Hence the party-state separation principle.

4.3.4 The Anti-faction Principle

We can now consider the final principle. We must accept that an elected democratic government is unlikely to represent all the people of a state at any one given time, where representation is understood in terms of voters’ electoral preferences as expressed on the ballot. But it does not follow that we should also accept that such an under-representative government only needs to serve the interests of those it represents. An under-representative government can, and should, still aspire to serve the interests of all its people. This is not drawn from an agenda for ensuring minority rights under a majoritarian government. Basic rights concern themselves with only the most fundamental human interests. Governments should never breach them unjustifiably. But governments do a lot more than not breach rights – when they work well, they facilitate human flourishing. Parties, in their idealised sense, work towards the flourishing of all the people of their state; factions care only for a sub-section thereof.⁹⁵ Factions a priori exclude the interests of their disfavoured section of the people from even being considered when framing policies – if these interests are considered at all, it is with a view to hurt them rather than to advance them. Importantly, given our capacity to threaten the very survival of humanity, at least in our times, factions would include parties that do not count the interests of the future people as legitimate concerns for their political calculations.

Factions fail to reduce the policy packaging costs for state institutions. We have seen that one of the key functions of political parties is to package the interests of all voters based on the party’s value commitments. These policy packages are then tested in elections, and voters express their preferences for or against such packages, which information is then available to state institutions when framing policy. In the process, parties also translate any voter’s factional interests into a subset of the common good through their policy packaging function, thereby moderating them to make them compatible with the interests of other citizens. Factions fail to do so. They also increase the political participation costs of the excluded voters – it is one thing to not have every party reflect one’s voter type, quite another to have a party in a system not even consider one’s interests as legitimate and relevant alongside the interests of all others.

This distinction between a party and a faction has been long recognised in political theory.⁹⁶ As Sartori put it, ‘If a party is not a party capable of governing for the sake of the whole, that is, in view of a general interest, then it does not differ from a faction. Although a party only represents a part, this part must take a non-partial approach to the whole.’⁹⁷ Factions are concerned with the interests and well-being of only a sub-section of the people. Parties, even when they make claims on behalf of particular groups, ‘must transcend the language of particularity and re-articulate the claims they represent in such a way that their demand for a share in political power is justified to the entire people and not only to that particular group of individuals that chooses to associate with them’.⁹⁸ The point of the distinction is normative rather than taxonomical: ‘[It] is very likely that the empirical analysis of existing practices will show how parties and factions are often entangled, with different political agents exhibiting features of both, to a greater or lesser extent.’⁹⁹ It is important to note that the distinction attaches itself to the entity as a whole and not to its individual actions. A party may have distinct policies catering to the interests of different sub-sections of the people – it will be a faction only if, taken as a whole, its political ideology and its policy platform is not justifiable to all the people. Any attempt to distinguish real-world parties from factions too sharply is likely to fail. Having said that, Rosenblum is probably right when she suggests that, even as an empirical matter, ‘where it is an original identity, or at least not reducible to prior political identities, the “we” of partisanship is more inclusive than other political identities’.¹⁰⁰

‘Rule of the people’ demands not only that political power is exercised by the people’s representatives but also that it is exercised in the name of all the people. In the words of White and Ypi, ‘the very ideal of collective self-rule implies that power is considered legitimate to the extent that it is justified to the whole people’.¹⁰¹ It is this normative ideal that leads us to our final principle: that a state should seek to ensure that political parties do not operate as factions. We will call this the ‘anti-faction principle’ and amend White and Ypi’s formulation somewhat to suggest that it requires political parties to ensure that their policies are objectively justifiable (rather than subjectively justified) to all the people. The amendment is required because it may be that a party fails to even communicate, let alone actually justify, its policies to all the people. So long as its policies are justifiable to all of them, the anti-faction principle should be satisfied. The anti-faction principle, therefore, does not require parties to articulate their policies in Rawlsian ‘public reason’ terms.¹⁰² Furthermore, a justifiability-standard is more tolerant of parties strategically appealing to particular sub-sections of the people as a matter of electoral tactics – so long as their packaged policy platforms are justifiable to all the people. Other independent moral and political constraints no doubt exist – such strategic appeals should not demonise any other section of the people, for example.

Unlike the purposive autonomy principle, which frowns upon single-issue parties, the anti-faction principle – on its own – does not require parties to have a plenary policy package. An anti-corruption party is not a faction. The party’s size doesn’t matter either, even if the party seeks only to influence policy rather than to govern. A small Green Party is likely to be a party, since its environmental objectives are justifiable to all. Even a party whose entire policy platform is devoted to advantaging a single societal group may not necessarily be a faction. A Workers’ Party, a Dalit Party in India (for former ‘untouchable’ castes), or an African American Party in the United States can be parties, if they can justify the interests of their preferred groups by reference to the general interest (for example, that historically excluded groups have a greater claim on the state’s resources). Furthermore, parties are allowed to make ideological and policy mistakes – the anti-faction principle does not demand that their policies actually work. But it does demand sincerity and plausibility – some obviously unworkable or implausible policies may evidence a lack of sincerity. A party that continues to deny the human impact on the global environment and its potential implications for future people, despite all the evidence to the contrary, is probably a faction because it is refusing to consider the interests of the future people, and its policies are unlikely to be justifiable to them. The one exception to the sincerity and plausibility test is this: even if a ‘party’ sincerely believes that the only interests that count are the interests of a sub-section of the people rather than those of all the people, its sincere rejection of democracy as rule-of-all-the-people is not enough to dodge its characterisation as a faction.

The hardest case is perhaps that of a party that claims that the sub-set of the people it represents is a distinct people – the Scottish National Party is a good example. Clearly, a full account of factionalism requires an understanding of who count as ‘a people’, as distinct from ‘a nation’ and ‘a citizenry’. What may be relevant to the determination is whether the party at least represents all the people in the constituencies in which it puts up candidates for elections. If it does, it is less likely to be a faction, if it doesn’t, more so. Perhaps the difficulty of this example demonstrates to us that it may well be parties, rather than constitutions, that constitute ‘a people’. The shared experience of a people as an audience for a party’s ideological and policy platform, in whose name and for whose benefit the party seeks to act, may well be constitutive. The failure of Europe to emerge as a robust polity may be owed to the absence of effective pan-European parties. This people-constituting feature of parties may well be the chief reason why factions are so problematic – they seek to redefine who the people that constitute a polity are by excluding a component thereof. Such redefinition may well be justified sometimes, such as when there is a legitimate right for internal or external self-determination. But outside this context, a majoritarian faction that excludes a minority from its concern is effectively seeking to exclude it from the polity by reconstituting its people on narrower, sectarian, terms.

The absence of a bright line dividing parties from factions may especially bother legal scholars: their fears are unfounded. I have already argued that apart from the application of anti-discrimination laws to parties, the anti-faction principle is best implemented by thorough second-order regulation: by creating a constitutional architecture – especially through the choice of an appropriate electoral system – which makes centrifugal factional behaviour politically less rewarding. Ranked-choice voting, cumulative voting, approval voting, and other such models, for example, are known especially to reward centripetal parties.¹⁰³ Such a regulatory architecture need never require a court to decide whether a party before it is a genuine party or a faction. This is precisely why the anti-faction principle demands that states ‘encourage’ anti-factionalism, unlike the party-state separation, which they ought to ‘ensure’. It is best implemented through political nudges rather than legal command-and-control. While the anti-faction principle may not be necessary for satisfying a wholly procedural conception of a democracy, only a slightly thicker conception of democracy is needed to appreciate that, for example, the capture of the ruling party/coalition by a small set of wealthy donors transforms a democracy into a plutocracy.¹⁰⁴ The principle is compatible with even an interest-bargaining model of democracy – so long as the interests of all social groups matter in the bargain; it does not necessarily require an endorsement of the more demanding republican or deliberative conceptions of democracy.

4.4 Conclusion

In this chapter, I have provided an idealised account of the functions of a political party in a healthy democracy that can help diagnose the ailments of unhealthy parties and party-systems. That account emphasises their Janus-faced role as intermediaries between the state and its people, which they perform by lowering key information and transaction costs in a democracy. Parties are therefore simultaneously public and private. Party systems that successfully reduce political participation costs, voters’ information costs, policy packaging costs, and ally prediction costs grease the wheels of representative democracy and are indispensable to its smooth operation. In order to aid parties in performing their intermediary function well, constitutions should seek to optimise four key principles in relation to political parties. First, they ought to protect the purposive autonomy of parties and align their rights and duties closely to their hybrid public-private character. Second, constitutions should optimise the number of parties such that there are enough parties to represent every salient voter-type, but not so many that voters’ information costs become unaffordable. Third, constitutions should ensure the separation of the parties from the state so that no party is able to entrench itself in the institutions and offices of the state. Breach of this principle increases the political participation costs of the supporters of opposition parties. Finally, the anti-faction principle requires that constitutions should encourage parties to cater to the interests of all the people, rather than merely of a sub-section thereof. Factional parties increase the political participation costs of excluded minorities. They also make policy packaging difficult.

Needless to say, these principles are compatible with many different ways of designing a constitution, but they are not compatible with all of them. How they apply precisely will obviously depend on the context. Note also that these principles may sometimes pull in opposite directions. For example, the anti-faction principle can be in tension with the party system optimality principle if a salient voter type demands factional representation. It is also in tension with the party state separation principle inasmuch as it demands that factions are structurally disadvantaged in the design of the state’s political institutions. The goal is therefore to attempt their collective optimisation rather than any one principle’s maximisation. In a separate article,¹⁰⁵ I have shown one set of institutional arrangements that might achieve such optimisation.

Political parties are the life-blood of representative democracy. Proposals seeking to respond to democratic deconsolidation have seen them as the problem, unnecessary middlemen who should be cut out of the system entirely in favour of direct democracy. These proposals fail to acknowledge the important function that parties perform in lowering key information and transaction costs that all-but-the-smallest democracies must contend with. If democracy is to survive, political parties need to be supported and improved, not eliminated. Despite all our advances in democratic technology, Schattschneider’s famous claim that ‘modern democracy is unthinkable save in terms of parties’ remains as true as it ever was.¹⁰⁶ Hence the four political principles that I argue should inform constitutional designs of democracies. A party-less democracy, unless really small, is unlikely to remain democratic for very long.