This article starts from an egalitarian perspective on fair representation in democracies. This perspective highlights three institutional requirements of fairness: (1) the electoral system should be proportional; (2) the decision rule should be majoritarian; and (3) the system of government should be parliamentary. A number of authors agree on these requirements (Christiano Reference Christiano1996; McGann Reference McGann2006; see also Bellamy Reference Bellamy2007). This normative perspective faces two important challenges.
First, while many authors agree on the three basic requirements, they disagree on other more specific aspects of an egalitarian democratic process. For instance, some authors who favour legislative majority rule justify a decision-making process in which parties make grand logrolls and bargains across different issues (McGann Reference McGann2006), whereas others argue that equality requires separable issues to be decided separately (Ward and Weale Reference Ward and Weale2010; Weale Reference Weale2013). Similarly, some proponents of proportional representation find it desirable that proportionally elected parties form pre-electoral coalitions (see Shugart Reference Shugart2001), whereas others consider such pre-electoral coalition-building to be normatively irrelevant or even undesirable (Christiano Reference Christiano1996; McGann Reference McGann2006). As these examples show, the egalitarian vision of democracy is under-determined; there are different ways to specify an egalitarian democratic process.
The second challenge is that the egalitarian vision of democracy may be too idealistic – too sanguine about the problems of holding those in power accountable. Political philosophers have formulated this challenge at a rather abstract level. For instance, Brennan and Pettit (Reference Brennan and Pettit2007: 261) doubt the feasibility of Christiano’s (Reference Christiano1996) ‘electoral-democratic vision of the polity’. In particular: ‘There are bound to be some in authority ... who will not perform as the ideal requires of them; they may pay more attention to ensuring their re-election, for example, than the ideal would condone’ (Brennan and Pettit Reference Brennan and Pettit2007: 261). In the empirical and comparative political science literature, a similar charge is framed in terms of an unavoidable trade-off between the fair and equal representation of voters and the accountability of governments (Powell Reference Powell2000). It is widely assumed that accountability is closely tied to ‘clarity of responsibility’ for legislative outcomes, so that some degree of disproportionality, and hence inequality, in the electoral system might be justified (Carey and Hix Reference Carey and Hix2011). Bol (Reference Bol2013: 3) summarizes this view nicely: ‘To increase the inclusiveness of representation, the cost of entry into parliament must be lowered, which decreases the chances of single-party governments being formed and undermines accountability and efficiency. This unavoidable trade-off typically creates contention.’ In short, we may have to reduce our egalitarian aspirations in order to achieve accountability.Footnote 1
What is often overlooked, though, is that important theoretical and empirical contributions raise doubts about the existence of a general trade-off between representation and accountability. For example, McGann (Reference McGann2013: 112) finds the conceptualization of accountability just sketched ‘downright perverse’, and Samuels and Hellwig (Reference Samuels and Hellwig2010) also criticize the idea that clarity of responsibility increases accountability for the socioeconomic condition. Based on a comprehensive empirical analysis, they claim that if one wants to facilitate strong forms of accountability, ‘one should push for institutional designs that generate low clarity of responsibility’ (Samuels and Hellwig Reference Samuels and Hellwig2010: 411, emphasis added).
This article addresses both challenges – the under-determination challenge and the trade-off challenge – simultaneously. It distinguishes four ideal-typical democratic visions based on the stages of the democratic process at which majorities are predominantly formed (Ganghof Reference Ganghof2015a; Ganghof et al. Reference Ganghof, Eppner and Heeß2015; Powell Reference Powell2000). Based on this typology, as well as on the comparative literature, three distinct versions of the egalitarian vision of democracy are reconstructed, all of which fulfil the egalitarian minimum requirements (proportionality, majority rule and parliamentarism). I contend that what is contentious between these visions is not the relative normative weights put on the values of equality and accountability. Instead, they are grounded in different conceptions of what the values of equality and accountability, as well as their reconciliation, require (cf. Gaus Reference Gaus2000). While some aspects of accountability are in conflict with certain conceptions of equality, the idea that there is a general trade-off between the two values is misleading.
The discussion proceeds as follows. First there is a discussion of the egalitarian perspective on fair representation. The next section then identifies four visions of majority formation in parliamentary systems, three of which meet the egalitarian requirement of proportional representation. The following two sections discuss the specific and controversial conceptions of equality and accountability assumed in these visions. Finally there is a brief conclusion.
Political Equality and Fair Representation
The time-honoured debate about fair representation is complicated by the fact that there are different conceptions of fairness. Hence relativism looms large: ‘The claim that there are many definitions of fairness ... slides very easily into the conclusion that they are all equally deserving of consideration’ (McGann Reference McGann2013: 90). McGann and others have tried to avoid this relativism by focusing on the basic democratic value of political equality. More precisely, the focus is on what he calls ‘liberal equality’: the idea that each citizen must be treated equally by the formal institutions of democracy. This is also a core idea for Christiano (Reference Christiano1996, Reference Christiano2008), who discusses and develops its deeper philosophical foundations. McGann (Reference McGann2013) insists that once we are committed to liberal equality – which we should be as democrats – we can understand certain institutional choices as logical requirements and thus in a quasi-objective manner.
Of course, the view that certain institutions are logically implied by liberal equality does not mean that we must always choose them. Both McGann and Christiano accept that there might be reasons for unequal institutions. Ganghof (Reference Ganghof2013, Reference Ganghof2015b) also highlights this point – following Gaus (Reference Gaus1996) – by advancing the notion of an equality presumption or baseline. The idea is that when thinking about the design of democracy, our starting point and default condition must be those institutions implied by liberal equality. The onus of justification is thus shifted to those who favour any inequality in the rules of democracy. There must be conclusive reasons for inequality, backed by systematic empirical evidence. If the reasons for inequality are inconclusive, so the argument goes, the democratic value of equality commits us to staying at the equality baseline.
Christiano (Reference Christiano1996) and McGann (Reference McGann2006, Reference McGann2013) have shown in detail how the combination of proportional representation (PR) and majority rule (MR) is implied by liberal equality and thus part of the equality baseline. This combination achieves a formally equal treatment between different alternatives and different voters. Each voter has an equal chance of selecting his or her favoured legislator or party; these fairly chosen legislators or parties have an equal chance of influencing the legislative outcome in parliament. The idea is to concentrate democratic decision-making power in a fairly constituted legislative majority.
Before we move on, it is useful to distinguish two definitions of proportionality: one is theoretical and applies to the mechanics of an electoral system, the other is empirical and applies to the outcome. The arguments of Christiano and McGann concern the mechanical understanding of proportionality: ‘An electoral system is proportional if it translates x percent of the vote into x percent of the seats of any party, real or hypothetical’ (McGann Reference McGann2006: 52, emphasis in the original). In contrast, the empirical definition of proportionality is based on a comparison of ‘the number of seats won by actual parties with the share of the votes won’ (McGann Reference McGann2006: 52). Mechanical proportionality is high when, as in the Netherlands, there is one nationwide electoral district and no legal threshold. In contrast, outcome proportionality can also be high in plurality elections, when ‘parties that would be severely underrepresented either do not run or people do not vote for them’ (McGann Reference McGann2006: 52). This distinction is important for the following discussion, and for understanding studies like that of Carey and Hix (Reference Carey and Hix2011). These authors argue that a limited reduction of (mechanical) proportionality is optimal, because it greatly increases one form of accountability and only moderately reduces (outcome) proportionality.
The argument for the final baseline requirement of liberal equality, a pure parliamentary system of government, can be seen as a straightforward extension of the reasoning by Christiano and McGann (Ganghof Reference Ganghof2015b). If power is to be concentrated in a fairly constituted majority, there must not be a chain of democratic legitimation that is separate from this majority. Hence direct elections of a president or prime minister are ruled out. Moreover, the fairly constituted majority must be the principal of the government and thus at any time be able to remove it. This requirement rules out executives that govern for fixed terms and without the possibility of a vote of no confidence. In other words, any further delegations of power should be within the authority of the fairly constituted majority. If it decides to delegate great powers to the executive – as modern large-scale democracies invariably do – then it must at all times have the power to dismiss the executive and select a new one. In this way the fairness inherent in how the legislative majority is constituted can be transferred to the executive. According to this logic, the selection or deselection of the executive are simply among the major decisions a fairly constituted legislature has to make. Hence pure parliamentarism (PP) is also part of the equality baseline, and any departure from it requires a special justification.Footnote 2 Our reasoning about the equality baseline can thus be summarized in a simple institutional design ‘formula’: Liberal equality=PR+MR+PP.
To be sure, there may be good reasons for departing from the baseline in some areas. For example, Christiano (Reference Christiano2008) argues that the value of (publicly visible) equality, which grounds the institutional design formula, also implies an extensive set of liberal rights (as well as an economic minimum). Therefore, if a majority violates liberal rights, it undercuts the very value on which its legitimate authority is based and undermines the moral rationale of the equality baseline. As a consequence, the nullification of laws by constitutional courts may be justified from an egalitarian perspective, even though courts are no representative institutions and imply a departure from majoritarian decision-making. We see here that by embracing an equality baseline, one does not necessarily subscribe to an unchecked form of democratic power – and one does certainly not deny the existence of trade-offs in the democratic design.
In what follows, I want to focus on the particular problem of trade-off mentioned in the introduction, that is, the potential trade-off between the (baseline) requirements of equal representation and the goal of government accountability. At the same time, I deal with the problem of under-determination, which is that the egalitarian baseline (PR+MR+PP) leaves more specific questions about democratic institutions and processes unanswered. I argue that to specify the desirable democratic process more fully, we need more specific conceptions of liberal equality, which are controversial even among authors who share an egalitarian perspective. The resulting visions of democracy also assume particular conceptions of accountability that are compatible with the chosen conception of equality. In spelling out these visions of democracy, I focus solely on party-based models of representation and ignore the question of how it can or should be combined with forms of personal representation.
Three Egalitarian Visions of Parliamentary Democracy
I want to start by distinguishing three visions of democratic representation that all satisfy the baseline combination (PR+MR+PP). To do so I follow Powell (Reference Powell2000) and focus on the main stages of the democratic process at which majorities can be formed (Figure 1).
Figure 1 Four Stages of Majority Formation
There are four main stages: the formation of parties, alliances, cabinets and individual pieces of legislation. I contend that each of these four stages corresponds to a distinct vision of democratic representation (Ganghof Reference Ganghof2015a; Ganghof et al. Reference Ganghof, Eppner and Heeß2015). Each vision is characterized by majorities predominantly being formed at one of the four stages. I briefly sketch all four visions, three of which are broadly consistent with the equality baseline. These three egalitarian visions are then elaborated on with respect to equality and accountability in the relevant sections of this article.
Party-centred Majority Formation
This is the vision of a pure two-party system, in which voters can directly select a one-party cabinet. Thus party formation is the crucial stage of majority formation. This vision is usually labelled ‘Westminster system’ or ‘majoritarian democracy’ (Lijphart Reference Lijphart2012; Powell Reference Powell2000). I do not expand on this model, as it is usually incompatible with proportional representation. Only under very unusual background conditions can a highly proportional electoral system generate a two-party system; Malta is a common example (Lijphart Reference Lijphart2012: 156). The party-centred vision plays an important indirect role in the following, though, as one prominent conception of accountability is derived from it. I shall also return to this vision later.
Alliance-centred Majority Formation
In this vision proportionally elected parties ideally group into two large pre-electoral coalitions or alliances (Powell Reference Powell2000: 12; Shugart Reference Shugart2001; see also Golder Reference Golder2006: 2–4, 137–8; Tillman Reference Tillman2013). To the extent that this is achieved, voters can fairly choose between parties, and they can at the same time directly select a cabinet – as in the case of party-centred majority formation. After the election the winning alliance takes office and its member parties recognize each other as veto players; that is, they commit to governing all decision-making issues together and forbid each other to enter into legislative coalitions with other parties.Footnote 3 Hence alliance formation becomes the crucial stage of majority formation. An important institutional feature of this vision is an electoral system that combines proportionality with significant incentives for the formation of two encompassing pre-electoral coalitions. Well-known examples are mixed-member proportional systems (Shugart Reference Shugart2001) and bonus-adjusted proportional representation systems (Bedock and Sauger Reference Bedock and Sauger2014). Germany in the 1980s and 1990s approximates this vision.
Cabinet-centred Majority Formation
In this vision, majority formation is postponed until after the election, but largely completed at the stage of cabinet formation. That is, proportionally elected parties compete independently, but they either form (multiparty) majority coalitions or minority cabinets with pre-negotiated and stable oppositional support. In both cases, the members of the legislative majority coalition also regard each other as veto players. Institutionally, this vision requires electoral systems that do not encourage pre-electoral alliance formation. Moreover, the rules governing legislative decision-making and executive–legislative relations should discourage the formation of minority cabinets without stable support. Demanding investiture rules are good examples (Bergman Reference Bergman1993). Finland approximates this vision of democracy (Ganghof et al. Reference Ganghof, Eppner and Heeß2015).
Legislature-centred Majority Formation
In the final vision majority formation is postponed until the formation of legislative majorities on specific issues or proposals. In pure parliamentary systems this usually requires the formation of (centrist) minority cabinets that seek flexible, issue-specific support in parliament (Strøm Reference Strøm1990). The oppositional support parties are not necessarily veto players in this case, and it is even possible that all or some cabinet parties are outvoted on some issues (Damgaard and Svensson Reference Damgaard and Svensson1989). The underlying ideal has been described as one in which the position of the median party is implemented on each separable dimension (Ward and Weale Reference Ward and Weale2010; Weale Reference Weale2013). Denmark has probably come closest to this vision of democracy (Christiansen and Damgaard Reference Christiansen and Damgaard2008).
Three Conceptions of Liberal Equality
The three proportional representation-based visions of majority formation – alliance-, cabinet- and legislature-centred – rely on different conceptions of liberal equality. These conceptions become more demanding as we move from earlier to later stages of majority formation (see Table 1).
Table 1 Three Egalitarian Visions of Democracy
Equality in the Alliance-centred Vision
This vision does not require more than the baseline demand of (mechanical) proportionality. Shugart (Reference Shugart2001) sketches this vision, although not from an explicitly egalitarian perspective. He argues that legislative majorities should be based on electoral majorities, which implies proportionality. No further requirements of fair representation are specified.
The conception of equality in the alliance-centred vision is constrained by its conception of accountability (elaborated on in the next section) in two ways. First, this conception focuses on the formation of pre-electoral coalitions between parties. Powell (Reference Powell2000: 53) suggests that if post-electoral coalitions are already formed before the election, there tend to be ‘tighter bonds between the parties’ and this ‘may include joint policy programs’. To the extent to which this programmatic integration of the pre-electoral alliances is achieved, however, individual parties’ ability to differentiate their programmatic stances – along one dimension or across separate dimensions – is necessarily reduced. Other analyses also suggest that the imperatives of forming pre-electoral coalitions tend to constrain the programmatic differentiation of parties within pre-electoral coalitions. Christiansen and Damgaard (Reference Christiansen and Damgaard2008: 69) make the qualitative observation for Scandinavia that ‘[p]re-electoral coalitions limit individual parties from pursuing an electoral strategy with a clear policy profile’. Similarly, Ganghof et al. (Reference Ganghof, Eppner and Heeß2015) present quantitative evidence for a negative effect of pre-electoral coalition-building on the multidimensionality of partisan preferences in advanced democracies. In short, unconstrained programmatic differentiation is not part of the equality conception in this vision of democracy.
Second, the alliance-centred vision may even require some degree of (mechanical) disproportionality and hence inequality in the electoral system. The reason is that as the effective number of parties increases, it is likely to become more difficult to include all of them into one of two alliances (due to the multidimensional differentiation of party positions, transaction costs, and so on). The conception of accountability in this vision may thus provide a justification for a limited departure from the equality baseline – although clearly a contestable one. An example is the 5 per cent threshold in Germany’s mixed-member proportional system, which is often seen as a prime example of the alliance vision (see Shugart Reference Shugart2001).
In sum, the alliance-centred vision of democracy has a very modest conception of equality – one that, at best, fulfils the minimum requirement of the equality baseline. The goal is merely to avoid overly strong institutional constraints on the effective number of parties.
Equality in the Cabinet-centred Vision
This vision goes one step further by also trying to avoid any restrictions on the differentiation of party programmes – both along a single dimension (polarization) and across separable dimensions (multidimensionality). While this programmatic differentiation is a problem in the alliance-centred vision – as just explained – it is of inherent normative value in the cabinet-centred vision (see Christiano Reference Christiano1996; McGann Reference McGann2006). It is seen as important in educating voters – especially the less knowledgeable ones – about how different issues are related to one another, and constraining it is considered an unnecessary simplification of the process of social discussion (Christiano Reference Christiano1996: 261).
So while the alliance-centred model contends itself with an unrestricted (or weakly restricted) number of parties, the cabinet-centred model also values an unrestricted programmatic differentiation of parties. It is interesting to note that the recent empirical literature also emphasizes this difference. Comparative studies show that the effective number of parties and their programmatic differentiation vary independently to some extent (Dalton Reference Dalton2008; Ganghof et al. Reference Ganghof, Eppner and Heeß2015; Singh Reference Singh2010). Democracies with similar effective numbers of parties can have very different levels of polarization and/or multidimensionality. There is also some evidence that the programmatic differentiation of parties may be more important than their effective number for explaining voters’ subjective feelings of being represented by a party or for their decision to vote in general elections (for example, Blais et al. Reference Blais, Singh and Dumitrescu2014; Dalton Reference Dalton2008; Wagner et al. Reference Wagner, Lehmann, Regel and Schultze2014). The different conceptions of equality in the two visions of democracy thus seem to matter empirically.
Equality in the Legislature-centred Vision
This vision embraces the first two goals – unrestricted number and programmatic differentiation of parties – but adds a third one. It recognizes that when proportionally elected parties form majority coalitions of veto players (as they do in the alliance- and cabinet-centred models), cabinet parties gain informal veto rights. These veto rights may be seen as a violation of the egalitarian baseline – as an informal departure from majority rule (Ganghof Reference Ganghof2015b). In this spirit, Ward and Weale (Reference Ward and Weale2010) suggest that the value of equality demands that informal minority vetoes should be avoided by separating issues as much as possible. Separable issues should be decided by separate majority votes in the legislature, so that no party becomes a veto player on all issues. In social choice terms, this would mean that on every separable issue a majority is formed around the median party (the party containing the median legislator), and policies are moved towards this party's position (cf. Black Reference Black1948). Ward and Weale (Reference Ward and Weale2010: 27) conclude that this sort of process has ‘unique normative properties, properties that should give it a special place in the theory of democracy. In particular, it is fair in the sense that it does not give a privileged role to any citizen or representative in the making of collective choices.’
The legislature-centred vision implicitly contains a critique of the alliance- and cabinet-centred visions – a critique that applies regardless of how the internal decision-making of fixed majority coalitions is modelled. There is a great deal of empirical disagreement about what the most adequate model is (Ganghof and Bräuninger Reference Ganghof and Bräuninger2006; Laver and Shepsle Reference Laver and Shepsle1996; McGann 2006; Martin and Vanberg Reference Martin and Vanberg2014; Tsebelis Reference Tsebelis2002; Warwick Reference Warwick2011), but most or all prominent models imply that the outcome of decision-making often diverges from the issue-specific median in parliament.
On the other hand, some of these models can also be used to criticize the legislature-centred vision. For example, if legislative parties are not purely policy-seeking (as assumed by Ward and Weale Reference Ward and Weale2010) but also vote- and office-seeking, oppositional support parties may also become veto players of sorts, and they can pull policies away from the issue-specific median (Ganghof and Bräuninger Reference Ganghof and Bräuninger2006). Hence the advantages of issue-specific majority formation may not be very large in practice. Second, one may question the normative attractiveness of achieving the issue-specific median. McGann (Reference McGann2006: 66–7) argues that ‘many policy choices are intrinsically multidimensional and interconnected’ and that ‘[a] great deal of policy-making is about the relative weight we give to different objectives’. Moreover, he contends that even when decision-making is multidimensional, policy will be moved into a relatively small, centrally located area of the policy space, the so-called ‘uncovered set’ (Miller Reference Miller2007). Parties (inside and outside the cabinet) still have to bargain over the exact location of policies within this central area, but this bargaining can be seen as desirable, as it can reveal important information about the intensity of preferences (McGann Reference McGann2006: 69). Parties or legislators can trade their support on issues they care little about for support on issues important to them (cf. Buchanan and Tullock Reference Buchanan and Tullock1962).
In sum, the three visions of majority formation differ significantly in how they spell out the requirements of political equality and fair representation (Table 1). They can all agree on the baseline requirements of proportional representation. The alliance-centred vision does not go beyond this requirement, and it may even require some degree of (mechanical) disproportionality. The cabinet-centred vision adds the requirement of the unconstrained differentiation of party programmes, and the legislation-centred vision adds the further requirement of unconstrained issue-separation in legislative decision-making. As already hinted, the differences in the demandingness of the equality conceptions are, at least in part, due to different conceptions of accountability. It is these conceptions that we turn to now.
Three Conceptions of Accountability
Even more so than equality, government accountability is a vague and contested concept that can be specified in different, even contradictory, ways. In the present context, a good starting point is Manin et al. (Reference Manin, Przeworski and Stokes1999: 40): ‘Governments are “accountable” if voters can discern whether governments are acting in their interest and sanction them appropriately.’ In other words, accountability involves seeing who are responsible and being able to sanction them. In what follows, I begin by distinguishing two broad approaches to accountability and then focus on how they are balanced and combined in the three visions of democracy.
Competing Approaches to Accountability
As noted in the introduction, one very influential approach to government accountability in comparative studies focuses strongly on the seeing aspect – that is, on ‘clarity of responsibility’ (Powell Reference Powell2000; Powell and Whitten Reference Powell and Whitten1993). This approach is largely derived from the Westminster model with its pure two-party system. The idea is that when there are only two parties, one in government and one in opposition, voters can clearly see which party is responsible and they can sanction the incumbent party by voting for the opposition. This perspective on accountability is often used to justify disproportional electoral systems. For example, Carey and Hix (Reference Carey and Hix2011) operationalize accountability in terms of two variables: the effective number of parties in the legislature (Laakso and Taagepera Reference Laakso and Taagepera1979) and the number of parties in government. Accountability is thus assumed to increase as the number of parties goes down and responsibility becomes clearer. They argue that proportional representation systems with low district magnitudes – median magnitudes between three and eight seats – are ‘best’ (Carey and Hix Reference Carey and Hix2011: 13) because they achieve a high accountability with relatively low ‘costs’ in terms of (outcome) disproportionality (Gallagher Reference Gallagher1991).
Prominent as the ‘clarity of responsibility’ approach is, however, it can be criticized as putting too much emphasis on the seeing aspect of accountability as well as on some mechanisms of sanctioning rather than others. One theoretical problem is that, if sanctioning means throwing the incumbent party out of office, then a disproportional electoral system may be counterproductive. Bad performance in office may lead the party to lose votes and seats, but disproportionality may help to keep it in office. Samuels and Hellwig (Reference Samuels and Hellwig2010: 400) give the example of the British Labour Party. Its vote share dropped from 41 per cent in 2001 to 35 per cent in 2005, which can be seen as sanctioning in response to a worsening economic situation. The party’s seat share dropped only from 63 to 55 per cent, though, so that it could remain in office. Hence the very same institutions that facilitate seeing them may also prevent effective sanctioning. To be sure, the British example is partly predicated on the fact that the Liberals, rather than the Conservatives, were the main beneficiary of the vote swing away from Labour. Their vote share increased from 18 to 22 per cent. In a purer two-party system, electoral disproportionality could have amplified the vote swing to cause a change of government. But the main point still stands: ‘clarity of responsibility’ as measured by single-party majority control is not necessarily correlated with accountability.
To move beyond examples, Samuels and Hellwig (Reference Samuels and Hellwig2010) present an instructive empirical analysis. They use three measures of accountability as their dependent variables, all of which are (with respect to parliamentary systems) focused on the prime minister’s party: seat share, vote share and control of government. They estimate how economic performance affects accountability, given a number of control variables. Performance is measured as the annual percentage change in capita GDP, either in the election year or the year before, depending on when the election was held. The results distinguish between ‘high’ and ‘low’ clarity of responsibility, based on whether a single party had majority control of the government. Their crucial result for parliamentary systems is as follows (Samuels and Hellwig Reference Samuels and Hellwig2010: 405, Table 1). In high-clarity situations, the state of the economy only had a statistically significant (positive) effect on one of the three measures of accountability: the incumbent party’s seat share. In low-clarity situations, in contrast, performance had a significant (positive) effect on all three measures, including the incumbent party’s ability to retain office.Footnote 4 Similar results were found for presidential and semi-presidential regimes. The authors thus conclude that ‘if one cares about the probability that incumbents survive in office as a function of socio-economic conditions ... then one should push for institutional designs that generate low clarity of responsibility’ (Samuels and Hellwig Reference Samuels and Hellwig2010: 411).
This critique of the ‘clarity of responsibility’ approach to accountability can be pushed one step further. We have said that throwing an incumbent one-party cabinet out of office is likely to be easier in a rather pure two-party system because the main opposition party is bound to profit from sanctioning the government. However, McGann (Reference McGann2013) notes that this might still not be an effective sanction, as the opposition party may have very similar positions on crucial issues (for example, on a decision to go to war). As in the previous critique, a voter might be able to see who is responsible but unable to sanction this party effectively. McGann thus sketches a very different conception of accountability, one that puts the focus on other causal mechanisms of sanctioning. The crucial condition for him is low entry barriers for new parties and thus the ability of voters to sanction all existing parties in the legislature. When entry barriers are low, voters can vote for new parties (without fear of wasting their vote) and thus reduce the seat and vote shares of the existing ones. According to this view, then, the central electoral requirement of the equality baseline, proportionality, also becomes a crucial requirement of accountability rather than being opposed to it: ‘Indeed to argue that a system that makes existing parties effective duopolists or oligopolists provides more accountability than one in which new parties can freely enter, seems to me downright perverse’ (McGann Reference McGann2013: 112).
My goal here is not to take sides in this debate but merely to show that the ‘clarity of responsibility’ approach is contested. Its proponents may object to Samuels and Hellwig that effective sanctioning requires not only a change of the party of the prime minister but a complete change of government – and that such complete alternation is more likely when the (effective) number of legislative parties is low (cf. Lundell Reference Lundell2011). In the same vein, one might object to McGann that new parties usually have fairly extreme positions on the left or right, so that – if they are not acceptable coalition partners – they may make a change of government more difficult. My point here is simply that different conceptions of accountability highlight different aspects and causal mechanisms, and that this fact belies the existence of a general trade-off between representation and accountability. In the following, I discuss how the three proportional representation-based visions of majority formation balance the different aspects of accountability (compare Table 1).
Accountability in the Alliance-centred Vision
At first sight, this vision can be seen as an effort to reconcile the two approaches to accountability: proportional representation facilitates effective sanctioning, while the formation of two competing pre-electoral coalitions facilitates seeing who is responsible. As to the latter point, Powell (Reference Powell2000: 53) presents a ranking of the ‘clarity of responsibility’ associated with different cabinet types. One-party majority cabinets are at the top, but majority governments based on pre-electoral coalitions come next. Powell’s (Reference Powell2000: 53) reasoning is that ‘the tighter bonds between the parties (which may include joint policy programs, candidates withdrawing in each other’s favour, and the like) make responsibility much clearer to voters.’ In other words, a crucial mechanism of improving the seeing aspect of accountability is the pre-electoral programmatic integration of parties – the lack of programmatic differentiation. This is not surprising, since the underlying conception of accountability is derived from the party-centred vision of majority formation. Hence accountability is increased if separate parties behave more like factions within a single party.
Yet this means that the two different approaches to accountability are not fully reconciled in this vision. We have already seen above that the successful formation of two competing pre-electoral coalitions probably requires a constraint on the number of legislative parties, and it also requires a constraint on the programmatic differentiation of these parties. Both of these constraints are directly opposed to the idea of unconstrained party entry. Even when new parties do gain representation despite (mechanical) disproportionality, they are unlikely to enter a pre-electoral alliance with established parties immediately. After all, new parties typically emerge out of protest against existing ones, or in order to emphasize issues that are not (according to them and their voters) adequately represented in parliament. Hence the idea of new party entry cannot be central to the alliance-centred vision. Clarity of responsibility is best achieved in a multiparty system, when the set of parties is fixed, whereas effective sanctioning of the existing parties may require the more or less continuous entry of new ones. The accountability conception of the alliance-centred vision can thus be summarized in terms of two goals: the pre-electoral programmatic integration of parties within the alliances and the post-electoral collective responsibility of the cabinet parties as veto players (Table 1).
Accountability in the Cabinet-centred Vision
It is arguably this vision of democracy that achieves some balance between the two approaches to accountability. Since it values the programmatic differentiation of parties for egalitarian reasons, it can also fully embrace purer proportionality and the free entry of parties as an accountability mechanism. In return, it has to give up the pre-electoral programmatic integration of parties. However, post-electoral collective responsibility of the cabinet parties is just as much possible as in the alliance-centred vision. Clarity of collective responsibility is clearly higher than in minority cabinets with shifting support (cf. Powell Reference Powell2000: 53). At the same time, relatively pure proportionality means that parties are not shielded by the electoral system against effective sanctioning. Voters can shift to existing opposition parties or to new parties without fearing that their votes are wasted. This accountability conception thus combines clarity of collective responsibility with unconstrained party entry (Table 1).
One objection to this idea of collective responsibility might be that voters do not tend to hold all cabinet (and declared support) parties equally to account but rather focus on the party of the prime minister (for example, Duch and Stevenson Reference Duch and Stevenson2008). While this is true to some extent, there is evidence for collective accountability of all cabinet parties, too (de Vries and Giger Reference de Vries and Giger2014). Moreover, there is also some evidence that the influence of cabinet parties on coalition compromises reflects their seat and portfolio shares (Martin and Vanberg Reference Martin and Vanberg2014; Warwick Reference Warwick2011), and that the party of the prime minister has advantages in intra-coalitional decision-making (Thomson et al. Reference Thomson, Royed, Naurin, Artés, Costello, Ennser-Jedenastik, Ferguson, Kostadinova, Moury, Pétry and Schermanne2014). Hence the attribution of somewhat greater responsibility to the prime minister’s party may be seen as a coherent element of this vision.
Accountability in the Legislature-centred Vision
The third vision retains the aspect of unconstrained party entry but sacrifices the clarity of collective responsibility. The reason is that this vision – based on its equality conception – aims for issue-specific decision-making on separable issues. Hence the collective form of accountability shared by the other two models is not possible. In the most extreme scenario, all parties in the legislature are part of the legislative majority on some issues and part of the minority on others. Hence, all of them or none of them is collectively responsible for the entire set of laws within some legislative period.
Proponents of this vision might argue, however, that the loss of collective accountability is to some extent compensated by better conditions for a form of party-specific accountability. While voters cannot clearly identify who is responsible for economic performance or other policy outcomes in general, they actually receive more information on which parties are members of particular legislative coalitions and thus responsible for particular decisions. Since parties are free to support some legislative proposals and reject others, responsibilities can be assigned in a more detailed manner. In contrast, the kind of grand logrolls required in the other two models – either before or after elections – are usually achieved behind closed doors, so that party-specific clarity of responsibility is reduced.
Of course, this idea of party-specific responsibility is very demanding, as it requires voters with fairly high levels of political sophistication. Recent research suggests that ‘the ability of voters to hold governments to account may be weaker than previously assumed’ and that ‘[o]nly highly sophisticated voters seem more able to use elections as an effective tool to . . . hold public officials accountable for past action’ (de Vries and Giger Reference de Vries and Giger2014: 357). This creates at least two problems. First, while in principle not many voters may be required to make party-specific responsibility work, it seems that as voters’ political sophistication increases, they become more ‘ideologically constrained’ and hence more resistant to switching parties (Dassonneville and Dejaeghere Reference Dassonneville and Dejaeghere2014: 595). Bluntly put, the voters who have the greatest capability of enforcing party-specific accountability may also have the lowest willingness to do so. Second, even if the sophisticated voters can enforce party-specific accountability, this is likely to exacerbate the inequality between them and the less sophisticated ones. Issue-specific decision-making may thus give sophisticates greater influence on political decisions. However, I am not aware of systematic empirical evidence on this question.
Discussion
Table 1 above summarizes the discussion:
1. The alliance-centred vision adopts a conception of accountability that is to a large extent derived from the Westminster model. This results in a modest conception of equality and may even (contestably) justify a moderate departure from the proportionality baseline. As a result, party entry is also constrained.
2. The cabinet-centred vision abandons the idea that programmatically integrated coalitions have to form before the election. Yet it can still achieve a form of collective cabinet accountability and combine it with unconstrained party entry. In addition, unconstrained proportionality and the absence of pre-electoral coalitions relate to a more demanding conception of equality – one that values the maximal differentiation of party programmes along and across issues.
3. The legislature-centred vision is based on an even more demanding conception of equality, one that highlights issue-specific majority formation. This has important consequences for the corresponding conception of accountability. There is unconstrained party entry but no collective accountability of a veto player coalition. Voters can hold specific parties accountable for their voting behaviour, but this is likely to be the prerogative of highly sophisticated voters.
We see a conceptual landscape that is very different from the standard view of a general representation-versus-accountability trade-off. Different conceptions of equality and accountability complement one another in the different visions of democracy. Indeed, we see that trade-offs between different mechanisms of accountability may be more important than trade-offs between equality and accountability.
Two clarifications are in order concerning this general conclusion. First, one may wonder whether I have, as it were, ‘defined away’ the general trade-off by keeping the party-centred vision mostly out of the analysis. This is not the case, because what we said about the other visions also applies to the party-centred vision: it is based on highly specific and contentious conceptions of equality and accountability. There are indeed well-developed normative accounts that see highly disproportional electoral systems such as first-past-the-post as being fully consistent with political equality (see, especially, Beitz Reference Beitz1989). And we have seen in the previous section that the party-centred vision puts great emphasis on some aspects of accountability at the expense of others, such as unconstrained party entry. Hence the party-centred vision, too, is not defined by the relative weight put on the values of equality and accountability, but by different ideas about what these values, as well as their reconciliation, require. It combines an even more reduced conception of formal political equality – one that relies merely on the minimal democratic requirement of equal voting rights – with an accountability conception that insists on one-party majority cabinets and two-party systems as a condition of true clarity of responsibility (cf. Powell Reference Powell2000: 53).
The reason why I have not included the party-centred vision in the above comparisons is that I do not want to contribute to the widespread relativism about fairness and equality. The work of Christiano (Reference Christiano1996), McGann (Reference McGann2006) and others has shown, in my view, that conceptualizing fairness purely in terms of equal voting rights is not demanding enough. The crucial point is the one that Christiano (Reference Christiano1996: 232–4) directs explicitly against Beitz (Reference Beitz1989). The latter evaluates the two main stages of the democratic process – the election of representatives and the passing of legislation by these representatives – separately, whereas Christiano and McGann evaluate them as an integrated whole. As Christiano (Reference Christiano1996: 233) puts it: ‘Political equality is not concerned with equality in determining who will be in the legislature but with equality in determining how the society will be arranged. The electoral process and the legislative system are components of this larger process; they are not end states of the process.’
In other words, what matters for ‘liberal equality’ is not that voters have an equal chance of getting some representative elected in some district (which they have in first-past-the-post systems, too), but that they have an equal chance of influencing legislation through their elected representatives. On this question I have indeed taken sides by arguing that full proportionality is part of the equality baseline.Footnote 5 This is why I have focused the comparison on the three proportional representation-based visions of democracy. My denial of a general trade-off between representation and accountability nevertheless applies to all four visions.
Since my focus has been on the realm of proportional representation-based visions, the argument is directed especially at studies that postulate the existence of some clearly ‘optimal’ system within that realm. Most notably, Shugart (Reference Shugart2001) suggests that the alliance-based vision is most ‘efficient’, and Carey and Hix (Reference Carey and Hix2011) argue that fairly disproportional proportional-representation systems optimally balance representation and accountability. The problem with these arguments is that they assume a consensus on the terms of the trade-off that does not exist. Since they rely on conceptions of representation and accountability derived from the party- and alliance-centred visions, they stack the decks against the cabinet- and legislature-centred visions from the start.Footnote 6
The second clarification concerns the scope of the present analysis. The goal here was to distinguish the four visions of equality and accountability at a basic conceptual level. We have not discussed the operationalization of the concepts in Table 1. Some of these concepts can be operationalized in different ways, and some may require more than one variable to represent them empirically, including variables that have not figured prominently in the above analysis for reasons of brevity. For example, Powell (Reference Powell2000: 61) emphasizes that a high level of cabinet stability is a ‘plausible condition’ for clarity of responsibility. So the fact that I have not discussed cabinet stability – or, more generally, the stability of the policy-making coalition – does not mean that I consider it unimportant. Elsewhere I have made suggestions about how to measure the four visions, which include cabinet stability (Ganghof Reference Ganghof2015a; Ganghof et al. Reference Ganghof, Eppner and Heeß2015).
Conclusion
We can distinguish three quite egalitarian visions of democracy, all of which fulfil the baseline requirements of proportional representation, majority rule and a parliamentary system of government. Keeping these three visions apart helps us to see that there is no general trade-off between equality and accountability. Instead, each vision is built on different assumptions about what these values require as well as how, and to what extent, they can be reconciled in practice. Conceptions of equality differ in their demandingness, and conceptions of accountability focus on different aspects of ‘seeing’ and ‘sanctioning’.
I want to make two suggestions for further research. First, the conceptual discussion here has focused on accountability and largely neglected the closely related idea of electoral mandates. This idea, too, has been closely tied to the party- and alliance-centred visions of democracy (Powell Reference Powell2000: 12; Shugart Reference Shugart2001). It might be worthwhile to spell out in more detail the conceptions of mandates underlying the different visions of democracy and connect them to empirical evidence on how faithful parties and governments are to their promises (for example, Thomson et al. Reference Thomson, Royed, Naurin, Artés, Costello, Ennser-Jedenastik, Ferguson, Kostadinova, Moury, Pétry and Schermanne2014).
Second, we have seen that the egalitarian baseline requirements do not go very far in ranking the alliance-, cabinet- and legislature-centred visions. Hence the comparative study of their performance is of major importance. I hope that the conceptual framework developed here can be useful in formulating new or refined hypotheses for empirical research. As to equality, for instance, we lack solid empirical knowledge about whether real-world minority cabinets with issue-specific support have the kind of normative advantages highlighted by Ward and Weale (Reference Ward and Weale2010). As to accountability, I do not know of any studies on the consequences of issue-specific decision-making for performance voting of (sophisticated and unsophisticated) voters. Similarly, I am not aware of studies that test whether majority cabinets based on pre-electoral coalitions are programmatically more integrated than others, or how such coalitions affect voter attribution of responsibility.
Testing these and other assumptions is important because we should ultimately evaluate the different visions of democracy on the basis of empirical evidence rather than normative intuitions. However, as long as a solid empirical basis is lacking, we must not forget that there are indeed competing conceptions of equal representation and government accountability. Ignoring this fact, and postulating a general trade-off between representation and accountability, can only complicate the debate about the institutional design of democracy.
