Richard S Katz. Handbook of Party Politics. Editor: Richard S Katz & William Crotty. 2006. Sage Publishing.
Notwithstanding that the democracy of Athens clearly predated the invention of political parties in the modern sense of formal organizations that promote candidates for office under a common identifying label, and notwithstanding the persistence of a few small democracies in which parties have not taken root (Anckar and Anckar, 2000) and the somewhat more common phenomenon of non-partisan local governments within systems that have parties at the national level, it is widely accepted ‘that the political parties created democracy and that modern democracy is unthinkable save in terms of the parties’ (Schattschneider, 1942: 1). Behind this apparent consensus, however, there is a wide range of views about what democracy means, and correspondingly about the proper nature and functions of political parties and party systems in a democracy.
Despite the great divergence of views, however, they can be organized around a relatively short list of interrelated questions. Three clusters of these questions bear directly on the definition of democracy:
- Is democracy primarily about the discovery and implementation of the ‘popular will,’ or is it primarily about popular imposition and enforcement of limits on government power? If democracy is about the implementation of the ‘popular will,’ how is that defined and identified? And if democracy is about the limitation of government, can this be reconciled with majority rule, or does it require that minority groups that might be victimized by an arbitrary majority be able to impose a veto?
- Is democracy primarily about outcomes or is it primarily about process, and if democracy is primarily about outcomes, is the meaning of ‘outcomes’ restricted to choices of policy or personnel, or does it extend to consequences for the moral or psychological development of citizens?
- Is democracy to be understood as a macro or a micro phenomenon—as Sartori (1965: 124) might ask, is ‘democracy on a large scale … the sum of many little democracies’? Sartori’s answer was no.).
- Others bear more directly on the nature of parties or party systems:
- Are parties properly considered as autonomous actors in the political process among which voters are called upon to choose or are they channels through which citizens themselves act politically?
- Ought parties to be distinctive and exclusive with regard to their policies and support bases, or ought they to be convergent and overlapping?
Parties in Popular Sovereignty Theories of Democracy
The simplest, and—at least through the 1970s -the dominant normative and (except with regard to the United States) empirical answers to these questions in Anglo-American political science begin with the ‘responsible two-party government’ model (Ranney 1962; American Political Science Association Committee on Political Parties, 1950; with regard to party government more generally, see Rose, 1974; Castles and Wildenmann, 1986; Katz, 1987). In its essence similar to Lijphart’s (1999) model of majoritarian democracy, which focuses heavily on political parties, but also to Riker’s (1982) model of populism in which parties are at most implicit, this model identifies democracy with the majority choice between two distinctive alternatives. These alternatives are embodied in political parties of which one, because there are exactly two, must win a majority at any election. That party then assumes control over the government until the next election, and because that control is undivided, the governing party can be held unequivocally responsible for its stewardship at the next election. Clearly this collective responsibility is contingent upon the cohesion of the party in power -and by extension to prospective voting, upon the cohesion of the opposition party as well.
Beyond this, however, the cohesion of the responsible two-party government model itself breaks down along a number of dimensions. The first concerns the basis of party cohesion. Is it cohesive support for a particular leader or team of leaders as in Beer’s model of Tory democracy (1982: 91-8), or for a particular line of policy, as in his model of Socialist democracy (1982: 79-86)? Alternatively, is electoral politics about finding the popular will by allowing the people to put ‘predominant political control in the possession of those who are by descent, by character, by education, and by experience best fitted to exercise it’ (Hearnshaw, 1933: 293-4) in a society where social classes are hierarchically ordered but not divided by fundamental interest (Tory democracy), or is electoral politics a peaceful alternative to warfare between classes whose interests are fundamentally opposed (Socialist democracy)?
Beer describes these models as ‘collectivist,’ but particularly the Socialist (policy-oriented) model has individualist equivalents in what I have described as binary and Downsian models of popular sovereignty (Katz, 1997). In the first, issues are assumed to form two clusters—the generalization of Duverger’s (1959: 215) claim ‘that political choice usually takes the form of a choice between two alternatives’—so that the two parties, each representing one of these clusters, take distinctly different policy stands. In the second, issue positions are assumed to be the equivalent of points on a policy line, with the parties remaining cohesive with respect to policy but tending to converge on the first preference of the median voter. The Downsian model of democracy also differs from binary democracy (and indeed from all of the other popular sovereignty models) in its assumptions (in the case of Downsian democracy explicit; in the cases of the other models implicit) concerning the relationship between the goals of policy-seeking and office-seeking (Strøm, 1990). In the Downsian model, parties formulate policy proposals in order to win elections; in the other models, parties try to win elections (seek office) in order to be able to formulate public policy.
The second dimension concerns the social basis of parties, in particular whether party is understood to be the political arm of a coherent social group or class in a fundamentally segmented society (e.g., the party of the working class or of farmers or of the religious) or alternatively as an alliance or representative of citizens who share common, but potentially mutable, views on issues. The former, which corresponds to the ‘i’ (ideological oppositions) end of the functional dimension developed by Lipset and Rokkan (1967: 10) from Parsons’ (1959) scheme of functional subsystems in society, is associated with the mass party of integration, and a political strategy of mobilization and encapsulation; the latter corresponds to the ‘a’ (interest-specific oppositions) end of the dimension, and to the catch-all (Kirchheimer, 1966), or electoral-professional (Panebianco, 1988: 262-7) models of party, and to a strategy of compromise and conversion.
The third, and related, dimension concerns what Ranney (1962: 156) described as ‘the little civil war about “internal democracy”’ On one side, early 20th-century scholars like Frank Goodnow (1900), the mid-20th century Committee on Political Parties of the American Political Science Association, advocates of the ‘Socialist democracy’ model of party government, and the parties of the ‘new left’ (although, as small parties, not advocates of the two-party model) see internal democracy as necessary for various combinations of three reasons. Particularly from the perspective of ‘Socialist democracy’ internal party democracy is essential in order for the party to be able to speak as the authentic voice of the social segment it represents, and this in turn both legitimizes the dominance of the party on the ground over the party in public office (Katz and Mair, 1993) and privileges the party manifesto over the individual judgement or consciences of elected officials. Particularly from the perspective of the new left, internal democracy allows ordinary party members to become actively involved as participants in policy-making for the party, and thus in true self-governance as citizens (see below). Finally, all advocates of internal party democracy see it as essential as a way to hold party leaders accountable, or, in the terms of principal-agent models, internal democracy is seen as a way of enforcing the control of party members as the principals over both party officials and public officials elected under the party’s banner as their agents.
On the other side, the principal argument is that internal democracy is incompatible with external cohesion (Ford, 1900, 1909; Downs, 1957: 25), or simply that democracy is about what happens between parties, not within them (Schattschneider, 1942). Moreover, if one assumes that the Downsian assumptions that lead to the expectation of party convergence on the first preference of the median voter would be translated to intraparty politics as well, then even if the parties could each present a coherent face to the electorate the result of intraparty democracy would be to fix each of the parties at the median of its own supporters, obviating the virtues of two-party competition. In principal-agent terms, the complaint is that public officials should be the agents of the electorate as a whole, and that the conflict of interest entailed in expecting them also to be the agents of their party membership organizations will allow, or even force, them to shirk this primary responsibility. Often in the American case this is supplemented by the claim that the entire enterprise—which in the absence of true membership organizations generally equates the party’s membership with its electoral supporters or at least its registrants or primary election voters (e.g., V.O. Key’s (1964: 163-5) category of ‘the party in the electorate’)—is fundamentally misguided: parties, in this view, are alliances of leaders between which voters choose, and not organizations of the citizens themselves (Schattschneider, 1942: 59).
The assumptions that there is a common interest that is identifiable by an elite who will be recognized by ordinary voters (Tory democracy), or that there are only two cohesive social groups competing for political power (Socialist democracy), or that even if there are two sides to every issue, they are the same two sides across all issues (binary democracy), or that an entire program of policy proposals can be reduced to a single point on a single dimension (Downsian democracy), each in its own way justifies a two-party system as both adequate and, given the presumed importance of having a stable majority outcome, desirable. Each is, however, also highly questionable.
Given its privileging of ‘strong and stable government first’ (Amery 1947: 19), there really is no Tory-like alternative to two-party government, and hence no way of dealing with the implausibility of its underlying assumptions about the nature of society and of the common interest or popular will. Each of the other models, however, can be adapted to a multi-party version that does not depend on such highly restrictive premises.
The easiest to adapt institutionally is Socialist democracy; if one replaces the Marxian assumption that there is an all-subsuming division between the proletarian and bourgeois classes with a more general cleavage-based view of society, then the Socialist prescription of two cohesive parties can be generalized to a prescription of a one-to-one correspondence between fundamental social segments and integrative parties. The problem is to get from this multi-party system of group representation to a decision process that can be assumed to result in the popular will, particularly in light of the well-known ‘paradox of voting’ (Brams, 1976; Condorcet, 1975 ; Arrow, 1963).
The same problem arises with individualist theories of popular sovereignty as soon as the binary or unidimensional assumptions are violated. One of the classic statements of the problem in the first case comes from Moisei Ostrogorski’s (1902: II, 618-19) study of late 19th-century politics in the United States:
What was pompously called the national verdict was, as a rule, tainted with ambiguity and uncertainty … after ‘the voice of the country had spoken,’ people did not know exactly what it had said; … for, however paramount a particular question may have been in the public mind, considerations foreign to it constantly entered into the ‘popular verdict.’
The problem in the second case has been extensively explored in the rational choice literature; simply, if preferences are not single-peaked (i.e., if there is not a single underlying dimension), then there is no stable equilibrium or Condorcet choice except under unbelievably restrictive conditions (Plott, 1967; for a full discussion of this literature in empirical as well as theoretical terms, see the chapter by Budge in this volume).
Ostrogorski’s solution was to replace permanent and all-encompassing parties with what would amount to a series of two-party systems (what he called ‘single issue parties’ (1902: II, 658-63), each of which would address one issue, resolve it, and then be replaced by a new pair of parties taking the two sides of a new issue.2 If issues are understood as continua rather than dichotomies, then one might imagine the same idea applied one dimension at a time instead of one issue at a time. Aside from the question of whether problems can be solved one-at-a-time and once-and-for-all, the obvious danger here is that if a succession of transient amateur administrations does not lead to chaos, it will lead to the dominance of the administrative officials as the sole possessors of the networks and expertise required for the government of a complex state (Ranney, 1962: 129).
The alternative is a multi-party system in which the parties are identified with unique combinations of policy positions rather than with social categories, and the problem of arriving at particular decisions on particular questions is transferred from the electoral arena to the legislature, leading to a model of democracy that might be called ‘legislative popular sovereignty.’ This opens the possibility of coalition formation through the trading of votes, and if not the likelihood of a Condorcet outcome, then at least the plausible argument that parties, as the representatives of groups of people who are in fundamental agreement across issues, and provided that their representation in the legislature is proportional to the size of those groups among the citizens, will reach the same outcome that the citizens would have reached themselves had direct democracy been possible.
Parties in Liberal Theories of Democracy
The six models of democracy suggested in the previous section all identify democracy with the discovery and implementation of the popular will. The major alternative family of democratic theories is concerned primarily with the liberal value of popular self-protection. This shift in value priorities implies a shift in attitude toward the relationship between government and citizens as well: for the popular sovereignty theories, government is a tool of the people, and therefore in a sense a part of them, whereas for the liberal theories the government is a potential danger to the people, and therefore necessarily separate. In principal-agent terms, while many, but by no means all, popular sovereignty theories recognize agency slack (with the individual parties or officeholders as the agents) as a potential problem, for liberal theories agency slack (with the government as the agent) is one of the two principal problems that must be addressed in institutionalizing democratic government.
All democratic theories must be concerned with the problem of the unchecked rule of a minority over the majority. That is, of course, the reason for the concern with majority formation in popular sovereignty theories. Liberal theories also rely on popular elections to control this danger. But where the converse problem of the unchecked rule of a minority by the majority is at most of secondary concern for popular sovereignty theories (which define democracy as the implementation of the will of the majority), it is the other principal problem for liberal democratic theories.
That said, liberal theories of democracy can be classified into four types, defined by the intersection of their assumptions about the nature of conflicting interests in society and therefore about the way in which what might be called majority tyranny can be avoided, on the one hand, and their assumptions about the relative commitments to liberal values of elites and of ordinary citizens, and therefore about the centrality of agency slack, on the other. As with the popular sovereignty theories, each set of assumptions has implications concerning the proper nature and role of political parties and party systems.
The first pair of liberal theories may be identified as ‘pluralist.’ The defining assumption is that the cleavages among interests are fluid and cross-cutting, rather than fixed and mutually reinforcing. As a result, institutions -including political parties—can be structured so as to make it difficult for any enduring majority to form while at the same time moderating the temptation of any majority of the moment to abuse its position.
Like the responsible two-party government models, these theories generally call for a two-party system, albeit primarily because this will allow the majority to evict a government -what Pinto-Duschinsky (1997, 1999) has called ‘removal van democracy’—rather than to ensure either stable government or popular endorsement of the policies to be pursued by government. They differ most centrally from the responsible two-parties models, and indeed in the American debate over responsible two-party government these pluralist liberal models represent the other side of the debate, in that they call for parties that are not cohesive with regard to policy, and especially not cohesive with regard to their social basis. Rather, the claim is that if each party depends at least in part on the support of groups that also support the other party, then neither party will be able to participate in the unreasonable exploitation of any group without itself suffering significant loss of support from that group.
This emphasis on weak and incoherent parties, coupled with confidence in the ability of ordinary voters armed simply with the right of frequent elections to prevent leaders from intruding on their rights, is particularly evident in pre-20th-century versions of majoritarian or pluralist liberal theory. Prominent examples include Jeremy Bentham’s call for annual elections to allow voters to ‘[divest] of their power all unfit representatives, before they had time to produce any lasting mischief (1962: III, 561) coupled with his hostility to political organizations, as exemplified by the weak cohesion of his own Radical Party, or the antipathy for political parties expressed in The Federalist coupled with the call for biennial elections of the House of Representatives.
Particularly after the collapse of democratic governments in Europe in the inter-war years, and in response to survey research that showed elites to be more committed to liberal values than were ordinary citizens, the focus shifted to place greater emphasis on the self-restraint of leaders, on the sociological preconditions of cross-cutting cleavages and dispersed and variegated access to politically relevant resources, and on regular competition among elites. Prominent examples of theorists of this genre include Schumpeter (1962), Sartori (1965), and Dahl (1956, 1966, 1971)—for a critique, see Bachrach (1967). As in the Downsian model of popular sovereignty, parties are understood to be teams of leaders rather than organizations of citizens, but unlike the Downsian assumption of prospective policy-oriented competition, the pluralist liberal ideal is retrospective result-oriented judgement by voters whose vocabulary is necessarily limited to ‘yes’ or ‘no’ (Schattschneider, 1942: 52). Moreover, because ordinary citizens are not only illiberal but also incompetent—‘the typical citizen drops down to a lower level of mental performance as soon as he enters the political field’ (Schumpeter, 1962: 262)—leaders ought not to be restrained (e.g., by intraparty democracy) from presenting the people with ‘results they never thought of and would not have approved of in advance’ (Schumpeter, 1962: 278).
If cross-cutting cleavages are a prerequisite for stable liberal democracy, as the pluralists argue, then such a regime ought not to exist in countries where cleavages are deep and mutually reinforcing. Yet, as Lijphart (1968) observed, the Netherlands (divided by religion), Belgium (divided by language), and Switzerland (divided by both religion and language) appeared to defy this rule. In response, he advanced the model of consociational democracy, based on his observation of democracy in the Netherlands. In many respects, this was a modern version of Calhoun’s (1943) model of ‘concurrent majorities,’ taking account of the fact that the fundamental social segments might not be coterminous with the geographic subdivisions of a federal state, and, as with the more recent versions of pluralist liberal theory, assuming that greater elite autonomy is both necessary and desirable. Together, they exemplify theories that can be identified as ‘veto-group liberalism.’
As the name implies, these theories solve the problem of majority exploitation by abandoning the majority principle altogether, and replacing it with a system of mutual vetoes. In Calhoun’s version of this model, the central government would be relatively limited in scope, and the vetoes would be exercised by the subnational governments. There is nothing directly said about the nature of the political parties operating within each subnational system, although given the tenor of his argument, it would appear that subnational governments are meant to operate under some version of responsible party government. One may infer, however, from Calhoun’s (1943: 34) disdain for separation of powers as an adequate guarantor against the majority faction, that coherent national parties are incompatible with his vision of democracy.
The basic claim of the consociational model is that ‘overarching cooperation at the elite level can be a substitute for crosscutting affiliations at the mass level’ (Lijphart, 1968: 200). It is a government by elite cartel characterized by four conditions: ‘government by a grand coalition of the political leaders of all significant segments of a plural society’; operation of a system of mutual vetoes by the leaders of all significant segments of society; proportionality as the standard for allocation of all or most political ‘goods’; and a high degree of internal autonomy for each group (Lijphart, 1968: 25). This system is supported by an electoral system of closed-list proportional representation with a single national district, all of which implies cohesive national political parties dominated by their leaders. Further, while the implication of Calhoun’s argument is that there would be a separate party system within each of the significant social segments (subnational units), in the ideal consociational system, there would be a one-to-one correspondence between parties and social groupings.7 Finally, in contrast to the pluralist vision of continuous, if moderated, interparty conflict between elections, the veto group liberal model assumes that electoral conflict will be replaced by inter-elite cooperation that is made necessary precisely by the mutual hostility of the social groupings’ members. This means that the parties’ leaders, whether initially chosen ‘democratically’ or not, must have the autonomy to override the unwillingness of their followers to compromise.
As indicated above, the responsible two-party government model corresponds quite closely to Lijphart’s model of majoritarian democracy. While the alternative, consensus democracy, has much in common with consociational democracy, from which it might be understood to be derived, it is also different in three important respects that bear on the role of political parties in democracy.
Consociational democracy explicitly rests on a system of mutual vetoes. In consensus democracy, the emphasis is instead on negotiation, and indeed Lijphart (1999: 2) indicates that Kaiser’s (1997: 434) term ‘negotiation democracy’ might be adopted as a synonym. With regard to the majority principle, which veto group liberalism overtly rejects as inadequate, consensus democracy takes a more ambiguous position. While oversized majorities clearly are preferred to minimum winning coalitions, this is still less than demanding a grand coalition. On the other side, minority governments are also preferred to minimum winning coalitions, because this implies constantly shifting majorities supporting particular issues in place of one stable majority. Nonetheless, the ultimate decision rule remains the majority principle. At the same time, however, the simple distinction between ‘the majority,’ which can be held accountable for all of the government’s decisions, and ‘the minority,’ which might replace it if enough voters change their preferences, is obscured.
The second respect in which consensus democracy differs from consociational democracy concerns the basis of the divisions among parties. In the consociational model, parties clearly are reflections of deep social cleavages. While the consensus model recognizes sociological dimensions of party competition, the very use of the dimensional metaphor, not to mention consideration of both a socioeconomic and a post-materialist dimension, implies a continuous multi-dimensional policy space in which parties choose positions, rather than a categoric ‘menu’ of parties directly corresponding to social structure.
The third difference concerns what I have called the ‘partyness of government’ (Katz, 1986: 40-6). While one might argue about which ‘aspect’ of a segment’s identity (party, church, press, union) is dominant in any particular arena, the fundamental point for the consociational model is that the leadership cadre of each segment forms a kind of ‘interlocking directorate’ (Lijphart, 1968: 59-70), so that party is intimately connected to the full range of governmental activity, and indeed the full range of political activity more widely understood. Among the defining characteristics of consensus democracy, on the other hand, is the explicit exclusion of party from the judiciary and central bank coupled with limitation of party influence through strong corporatist institutions (which are understood to be competitors for party dominance), territorial division of power, and presidentialism.
The conjunction of these three differences implies a model that is different both from popular sovereignty, which is about finding the popular will, and from liberalism, which is about the containment of conflict. Instead, consensus democracy appears to be about equitable management in the pursuit of objectives that have been defined a priori to be good. One of these ‘goods’ is demographic representativeness. Having cut party free from social cleavage, and in any event including gender—which has never been taken to define a social cleavage analogous to race or language group, or economic sector—among the traits that ought to be mirrored in a representative body, one must presume that parties individually are to be demographically representative, either of the population as a whole or of their individual but heterogeneous electorates. Another ‘good’ is economic prudence; presumably parties should both advocate economically prudent policies and advance leaders who will be competent in their pursuit. In other words, consensus democracy appears to limit not only the role of parties in societal governance, but also the scope for politics within the parties. Indeed, in its tendency to judge outcomes by technocratic standards, one might ask whether it also limits the scope for citizens to engage in politics, and in this sense ask whether it is an alternative form of democracy, or rather a lesser democracy.
Democracy as a Way of Life
Both popular sovereignty and liberal theories define democracy in terms of what the government does (or does not do). While they often make prescriptions concerning process, these are justified by their impact on, or necessity for, the valued policy outputs. Other theories, however, identify democracy with its impact on those who participate in it, which they attribute to the process itself. In contrast to Schumpeter’s (1962: 242) definition of democracy as a ‘type of institutional arrangement for arriving at political—legislative and administrative—decisions,’ these theories in their fullest form see democracy as a way of life. As John Dewey (1927: 143) put it, ‘The idea of democracy is a wider and fuller idea than can be exemplified in the state even at its best. To be realized it must affect all modes of human association, the family, the school, industry, religion.’
Even if attention is limited to the narrowly political sphere, in which political parties may be expected to be relevant, these theories are concerned with the development of citizens in the sense of being efficacious people who have fully developed their individual capacities and have achieved self-mastery (Bachrach, 1967: 4; Mill, 1962: 49-52; Lane, 1962: 161 (quoting Margaret Mead)) and of understanding themselves to be members of a community who can and do make civil judgements and evaluate goods in public terms (Mill, 1962: 71-3; Barber, 1984: 158), rather than merely being citizens in the juridical sense. Clearly simply going into a voting booth every few years and making a private choice (both in the sense of being motivated by private passions and in the sense of neither being revealed nor justified in public) among parties or candidates will be inadequate to further these objectives. But what is the role of parties in this process of democratic development?
One answer has already been indicated with regard to the attitudes of new left parties to intra-party democracy (Kitschelt, 1982). Particularly in their local branches, parties can provide a venue in which citizens can be actively engaged in the actual doing of politics: debating issues (rather than merely listening to debates) and making decisions (rather than merely voting for others who will make decisions for them). This is a more extensive and continuous democracy than the representative model of the mass party with its congress and national executive committee, or the popular choice of candidates through a direct primary. Organizationally, it calls for frequent meetings, decision by members rather than delegates, and rotation of office-holders both within the party and among the party’s elected officials. The contrast between this understanding of intraparty democracy, and thus of the proper nature of parties themselves, and the understanding in the responsible party government models is illustrated by the debate within the German Green party between the ‘Realos’ and the ‘Fundis.’ For the Realos, the possibility of policy influence in the Bundestag or even coalition membership required abandonment of rotation of office-holders, and granting party leaders the authority to reach accommodations with other parties; for the Fundis, the principles of democratic life within the party were more important than influencing government decisions more directly.
While all this may be possible in a small party and with regard to local questions, a large or geographically dispersed party will require more formal organization and representative institutions. This raises the problem of the ‘iron law of oligarchy,’ posited by Michels (1962) on the basis of his study of the German Social Democratic Party—a party ideologically committed to democracy, but in Michels’ view profoundly undemocratic in its actual functioning. In Michels’ (1962: 365) view, the problems that he observed were the result of organization: ‘[I]t is organization which gives birth to the dominion of the elected over the electors, of the mandataries over the mandators, of the delegates over the delegators. Who says organization says oligarchy.’ The precise implication for the relationship between parties and democracy is open to debate. Most commonly, the iron law is taken to imply that large-scale democracy is impossible, and that if political parties are essential institutions of large-scale democracy, then they are part of the problem. If democracy is a state to be achieved, then this reading clearly is correct. If, however, democracy is a state to be approached to a greater or lesser degree, then Michels himself suggests an alternative interpretation. Within only a few pages of the iron law itself, Michels (1962: 366, 369) observes that only the ‘blind and fanatical’ could fail ‘to see that the democratic current daily makes undeniable advance.’ Moreover, ‘[s]ometimes … the democratic principle carries with it, if not a cure, at least a palliative, for the disease of oligarchy.’ A democratic movement (or society), ‘in virtue of the theoretical postulates it proclaims, is apt to bring into existence (in opposition to the will of the leaders) a certain number of free spirits who … desire to revise the base upon which authority is established.’ If all this is true, then although party as organization may contribute to the problem, party as the carrier of the democratic ideology may be a palliative.
In the American context, Herbert Croly (1909, 1914) also argued that political parties, whatever their virtues in the original transition from colonial rule, were an impediment to democracy. The reason was twofold. On one hand, parties imply ‘a separation of actual political power from official political responsibility … The leader or leaders of the [electoral] machine are the rulers of the community, even though they occupy no offices and cannot be held in any way publicly responsible’ (Croly, 1909: 125). On the other hand, parties demand of citizens ‘that they think and act in politics not under the influence of their natural class or personal convictions, but according to the necessities of an artificial partisan classification’ (Croly, 1914: 341). Democracy, however, requires direct rather than mediated public decision: initiative, referendum, and recall, rather than periodic partisan election (Croly, 1914: 324). It also requires citizens to think and act as members of a community, habits which party, precisely because it is ‘part,’ destroys. His conclusion was that democracy was, at least when he was writing, impossible at the national level in the United States, where representative government, which could never be truly democratic in part because it requires parties, would have to do. At the state level, however, institutions could be devised that would allow what he identified as direct democracy. Democracy could thus be advanced by shifting the locus of power to smaller units, which would make parties unnecessary.
This prescription, that power be shifted to more local units, is a recurring theme among democratic theorists who are more concerned with the impact of democracy on the human development of citizens than with the policy outputs of government (Pateman, 1970; Barber, 1984; Mansbridge, 1980). A second recurring theme is the importance of applying the democratic virtues of equality and collective self-rule beyond the narrowly governmental, in particular to include workplace democracy. What is conspicuous by its marginality and often by its complete absence, is any mention of political parties.
Parties and Representation
No discussion of contemporary democracy can ignore the fact that modern democracy necessarily is representative democracy. And since parties are intimately involved in the process of representation, this means that consideration of the place of parties in democratic theory must address the place of parties in the theory of representation.
Analysis of representation involves three questions: Who are the represented? Who is the representative? What is it that the representative does in representing the represented?
Beginning with the last of these questions, the literature suggests five basic answers. The first mode of representation is the descriptive mirroring of demographic characteristics. With the second mode of representation, it is the distribution of opinions rather than of personal characteristics that is to be mirrored. With these two modes, representation means standing for the represented. The other three modes understand representation as acting for the represented. The third and fourth modes relate to the classic distinction between the representative as delegate and the representative as trustee (see, for example, Wahlke et al., 1962). The delegate serves as the direct agent of his or her constituents, doing what the represented want him or her to do, serving as a conduit for their opinions, following their direct instructions. The trustee, on the other hand, acts for the represented by using his or her own judgement to advance their interests, but not necessarily in accordance with their currently expressed opinions. The fifth mode of representation is to act for the represented in the role of ombudsman, or more generally of provider of constituency service.
While party is implicated as the representative in all five of these modes of representation, whether it is the individual but partisan official, the constituency party, the national party, or the parliament as a whole (with the individual parties contributing to its composition) that should be understood as the primary representative varies from one mode of representation to another, and among alternative conceptions of democracy. Similarly, whether the represented should be understood primarily as the citizenry as a whole, the citizens of particular areas or groups, the citizens who are voters of the party (again either as a whole or in particular areas or groups), the individual citizens, or indeed the party membership organization itself also varies depending on the particular sense of representation and the general conception of democracy being considered.
In its original implications, the distinction between trustee and delegate roles refers not just to the decision process underlying the representative’s vote (do what one’s constituents want or do what they would want if they were as wise as the representative), but to the nature of the democratic process as well. The delegate orientation only makes sense if one regards democracy primarily to be about the aggregation of interests or opinions; since the citizens cannot all be present to express their views or defend their interests, they ‘hire’ a representative to speak for them. If, however, democracy is about deliberation, then one naturally asks, as Edmund Burke asked in his 1774 address to the electors of Bristol, ‘what sort of reason is that, in which the determination precedes the discussion; in which one set of men deliberate, and another decide; and where those who form the conclusion are perhaps three hundred miles distant from those who hear the arguments?’ and thus arrives at a trusteeship model of representation. Parties contribute to this kind of representation as the members of parliament, but it is a form of representation in which the real representative is the parliament as a whole representing the people as a whole. This is very much an 18th-century view of representation based on an assumption of nascent and weak parties, but it is also reflected in the model of consensus democracy, with its emphasis on parliamentary (as opposed to cabinet) power, and Tory democracy, with its emphasis on independent leadership pursuing a singular national interest.
The conjunction of the idea of parliament as the representative and the idea that deliberation is essential to democracy also underlies many of the claims for representation in the sense of mirroring. The claim that the demographic characteristics of the citizens should be mirrored by the representatives has three primary justifications: that inclusion of minority or otherwise disadvantaged groups signifies their status as full citizens; that people in different social positions bring different experiences to bear on deliberations; and that differing groups have interests that are sufficiently at variance that a member of one group cannot represent the interests of another (see Kymlicka, 1993: 67, and the works cited there). The claim for the mirroring of opinions, on the other hand, has two justifications. The first is instrumental, and has already been elaborated in the discussion of legislative democracy: since there are too many possible combinations of policies across diverse issues for any one combination to receive majority support in the electorate in competition with all of the others, and since it is impossible to anticipate all of the questions that might arise between elections, parliament can only be expected to make the decisions that the people would have made themselves if it reflects the full diversity of their views. While this is true even if decision-making is simply the serial taking of votes on isolated issues, it is especially true if one takes into account alternative possibilities for vote trading or compromise and accommodation, both of which could be understood as equivalent to deliberation. The other justification for the mirroring of opinions is expressive: those who hold unpopular views deserve to have those views expressed in parliament, regardless of outcome, and moreover are more likely peacefully to accept their defeat if they have, at least, had their say.
The equality of citizens is a vital value to any theory of democracy. To say that demographic mirroring is important for its symbolic attestation of equality, however, is to imply that it is representation that would not happen otherwise. It thus suggests that parties should take affirmative steps, such as the implementation of ethnic or gender quotas both for positions within their own organizations and in the selection of candidates for public office. On the other hand, precisely because this form of representation is symbolic, it has also been associated with what might be called ‘sham democracies’; the high demographic representativeness of the parliaments of Soviet bloc countries was indicative of the powerlessness of those bodies—and correspondingly of the impotence of political parties as well.
The idea of demographic mirroring, or at least the direct representation of members of traditionally disadvantaged groups, has recently been recast in terms of a ‘politics of presence’ (Phillips, 1995). While retaining some of the symbolic argument (it is harder to treat a group unfairly when some of its members are in the room), this is supplemented by the claim that the lived experience of members of these groups gives them a perspective that cannot be represented adequately by a mere sympathizer. This has two important implications for parties and representation. First, it clearly implies a deliberative assembly and therefore a trusteeship model of representation. Second, if parties are to be significant at all, then it calls for descriptive representation not only in parliament as a whole, but within each party, and within the executive, as well. (See the chapter by Paul Lewis in this volume.)
Obviously representation as mirroring can only be effected by a collective representative, but should the individual parties be representative (either of their own electorates or memberships, or each representative of the population as whole), or is it only the parliament that must be representative (which would naturally result from the aggregation of individually representative parties, but might also occur even if the individual parties were unrepresentative)—or must not only the parliament but also the cabinet be demographically representative? While these questions have been raised in the literature primarily with respect to judging electoral systems, they also have implications for parties. On one hand, mirroring by the parliament as a whole may have little practical value if the true locus of deliberation and decision is the majority party caucus or the cabinet room. On the other hand, the more parties are constrained to look and think like the population as a whole, the less substantive choice is left to the voters.
The emphasis on deliberation that is implicit in representation as demographic mirroring, and indeed in representation as opinion mirroring, raises another tension in the theory of representative democracy. Are elections primarily about the choice of representatives, or are they primarily about the choice of government (Milnor, 1969)? In part, to address this question is simply to revisit the question of the appropriateness of two-party versus multiparty systems raised with regard to popular sovereignty and liberal models of democracy. In part, it revisits the question of whether the venue in which representation primarily takes place and should be assessed is the parliament as a whole (choice of representatives) or the decision-making venue of the governing party or coalition (choice of government). It also raises the question of delegation or trusteeship, since the negotiators of post-election coalition agreements cannot in any strong sense be other than trustees of the voters who supported their parties, and indeed, even with the greater possibility of consultation during the negotiating process, must largely operate as trustees of their party organizations as well.
This overview of the question of parties in democratic theory has necessarily been incomplete. As the careful reader will have noted, it has also left a number of loose ends. Some of these are tied up (or at least tied off) in other chapters of this Handbook, but many are not. The careful reader also will have noted sections with few or no references to ‘the literature.’ Both the loose ends and the scarce references reflect the facts that although the literature of democratic theory is immense, it has developed largely without reference to the richness and complexity of empirical studies of political parties, and that although scholars of parties often make introductory reference to their centrality to modern democracy, they rarely go beyond this to consider the distinctions among varieties of normative democratic theories. When parties scholars consider varieties of democracy, they usually refer to the distinctions between presidential and parliamentary systems, or between bipolar and fragmented patterns of competition. When democratic theorists think about parties, it is generally to ask whether they need be democratic in their own organizations, or indeed whether parties are a precondition or an impediment to democracy.
Work that took seriously both the empirical study of parties and the normative complexity of democracy tended to appear in the period between about 1880 and 1920, as the transition from régimes censitaires to mass suffrage democracy was taking place, with the attendant transformation from cadre to mass parties, and again in the period between 1945 and 1960, in response to the problems of reestablishing democracy in Germany and Italy and attempting to establish democracy in the former colonies of the British and French empires. With the collapse of communist regimes in eastern and central Europe, it would appear that the stage is set for a third wave of analyses connecting parties and democracy.