Jump to ContentJump to Main Navigation
Elections, Parties, DemocracyConferring the Median Mandate$

Michael D. McDonald and Ian Budge

Print publication date: 2005

Print ISBN-13: 9780199286720

Published to Oxford Scholarship Online: February 2006

DOI: 10.1093/0199286728.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.  Subscriber: null; date: 17 September 2019

Choosing Governments or Identifying Preferences? The Role of Elections in Democracy

Choosing Governments or Identifying Preferences? The Role of Elections in Democracy

Chapter:
(p.3) 1 Choosing Governments or Identifying Preferences? The Role of Elections in Democracy
Source:
Elections, Parties, Democracy
Author(s):

Michael D. McDonald (Contributor Webpage)

Ian Budge (Contributor Webpage)

Publisher:
Oxford University Press
DOI:10.1093/0199286728.003.0001

Abstract and Keywords

This chapter lays out the general questions as described in the overview. Is it really satisfactory to have ‘democracy with qualifiers’ (majoritarian, consensus, and so on) each justified in their own terms? A unifying median mandate approach is proposed which also gives a guarantee of a necessary connection between popular preferences and public policy, which no other account of representative democracy provides.

Keywords:   democracy, government mandate, median mandate, popular preferences, public policy, elections, political parties, representative democracy, majoritarian democracy

1.1 ELECTIONS AND GOVERNANCE

Elections are the distinguishing institution of democracy, translating individual voter preferences into collective choices that can in some sense be said to reflect them. The key questions about democracy therefore focus on elections, starting with the aggregation process itself. How are preferences reflected in votes? How are these combined to produce the overall election result? Once the election result is declared, what is its relationship, if any, with public policy? Does it determine it, in any way, or does it simply produce a parliament and government that then make policy themselves?

Classic theories of representation are in no doubt about this. They see democratic choice as a two‐stage process. Voters choose legislatures and governments, which then, autonomously, make decisions for them. This is not dissimilar to ‘consociationalist’ or ‘consensus’ conceptions of democracy (Lijphart 1984, 1999), where party leaders negotiate policy compromises that deeply divided populations cannot agree on. In both cases the role of elections is to choose legislatures, which then independently shape both governments and policy.

Still building on the idea of a two‐stage process, but shifting the focus from legislatures to government, current mandate theory sees parties as offering alternative policy programmes to electors. The most popular programme attracts majority endorsement and propels its sponsor into government with majority backing. The ensuing ‘elective dictatorship’ then carries through the promised programme, for which it is held accountable at the next election.

Mandate theory offers a way of coming to terms with the dominance of political parties within representative structures. As a result, it has guided most contemporary theorizing about, and most empirical investigations of, the way democracy works. From Downsian spatial modelling (Downs 1957) to theories of coalition formation (Laver and Budge 1992) the emphasis has been on how parties form governments.

(p.4) The problem for government mandate theory once it moves into the real world is that only one democracy in the world—the USA—regularly produces a spontaneous majority choice of governing party. Elsewhere majorities are manufactured, if they are produced at all, either through the mechanics of the electoral system or by negotiations between potential coalition partners. This has led much research on democracy to concentrate on government formation as the key process in its functioning. Parties are carriers of specific policies (Castles 1982; Budge and Keman 1990). So once government composition is settled the policy mix to be adopted as public policy will follow automatically.

How exactly this relates to majority preferences, however, is obscure. This obscurity has been seized upon by ‘consensus democrats’, who refuse to be thrown by the dubiously majoritarian status of most governments. Going back to earlier ‘trustee’ conceptions of representation, they argue that the authority conferred by election enables party leaders, whether in government or in opposition, to negotiate necessary compromises, which they can then persuade voters to accept. This conception is not unrelated to the brokerage role assigned by Madison (1788/1911) and later American writers to policy bargaining within and between political institutions that goes on autonomously from electors.

The problem in both cases is the absence of any institutional mechanism ensuring ‘a necessary correspondence between acts of governance and the equally weighted felt interests of citizens with respect to these acts’ (May 1978; Saward 1998: 51; cf. Weale 1999: 14). Representatives and parties may be benevolent and consensual and consider the general good. But there is no mechanism to make them do so. They could thus be partisan, narrowly self‐interested, and successfully manipulative and still get re‐elected, especially given difficulties with accountability in coalition situations (Powell and Whitten 1993). As normative theorists recognize, what distinguishes democracy from benevolent despotism are precisely its institutional mechanisms for ensuring a necessary correspondence between government policy and individual preferences, not one dependent on vagaries of culture, elite temperament, or the goodwill of rulers.

More difficulties of a government‐focused view of democracy are considered in detail in later chapters, both theoretically and with comparative data. These analyses point strongly to a radical solution for the conceptual and practical ambiguities associated with the traditional approach to elections—that is, abandoning the idea that elections really determine which parties govern. We can see them instead as specifying the policy preference of the popular majority. Operationally this is done by locating the position of the median or middle elector. With near 50 per cent of other electors to one side and near 50 per cent to the other, the median elector has the decisive role in forming a popular majority that must adopt his or her (p.5) position—otherwise he or she will either abstain or vote with the other side. The power to force the median preference on the majority is desirable, not only because it provides an anchor point for majority opinion in the first place but also because it is the endorsed policy position that minimizes differences with all the others, given equal weighting of votes.1 It thus provides everyone with the best they can get in public policy terms under the existing distribution of individual preferences.

Elections provide a mechanism for identifying the median preferences from the distribution of votes over the various policy alternatives offered by parties. They also indicate which party is its carrier (or comes nearest to being so). The electoral system should, and often does, ensure that this party also contains the median member of the legislature—determining of public policy therefore just as the median elector is of the popular majority and for the same reason. No legislative majority can be formed without that party.

It is important to recognize that identification and empowerment of the median position constitute a logical extension of traditional party mandate ideas rather than a replacement or a rival to them. As we shall see in Chapter 2, both ‘median mandate’ and ‘government mandate’ base themselves on much the same set of assumptions about party and electoral behaviour. The sole difference lies in the idea that the mandate is given to a single‐party government by a cohesive popular majority. Median mandate theory recognizes that such a majority will exist only rarely and looks for an acceptable substitute in the other cases. For the reasons given above, it finds the substitute in the median voter position that then mandates its party carrier to effect it in office. At the extreme, where there is a popular majority, the two versions of the mandate merge, for this is the special case where the median voter is found in the majority anyway. All that median mandate ideas reject from the traditional mandate approach is its tendency to compromise with practice by endowing mere pluralities with the attributes of a genuine majority. That is unacceptable, as a plurality‐based government is actually opposed by the majority of voters unless it is at the median.

These consequences depend of course on there actually being a median policy position. As we shall see this is facilitated by, but by no means dependent on, opinion being ranged along a single dimension of policy, usually a Left‐Right one. We discuss the dimensionality of policy spaces in Chapter 3 both theoretically and with evidence from post‐war democracies. Elections do seem to have the property of compressing political differences into a single set of Left‐Right differences for the duration of the campaign and a brief time thereafter, however fragmented they become for most of the (p.6) interelection period. Even a fragmented interelection policy agenda however usually groups left‐wing parties against right‐wing parties on most central issues.

Its strategic position does of course make the median party a natural member of governments. In the extreme case of a party that gains a majority of votes, it will form the government itself with a direct mandate to carry through its policies. Coalitions tend to include median parties—in about 80 per cent of post‐war governments (according to Laver and Budge 1992: 416). For the reasons given above such parties form a natural anchor point for the parties around them. Being in government naturally reinforces the ability of the median party to get its policy accepted. This is also, under appropriate election arrangements, the policy preferred by the majority of electors once they have constituted themselves as a majority.

We have not arrived back at theories of elections as determining government composition, and through that public policy, for two reasons:

  1. 1. Policy payoffs from being in government are not clear in many cases (Laver and Budge 1992: 423–5).

  2. 2. Twenty per cent of coalition governments do not include the median party. Unless we are to write the latter off as a negation of democracy, we have to recognize that a median party, representing the majority preference of the electorate, can influence policy in these cases too and for the same reasons. A legislative majority, whether narrow or wide, has to include the median party or at least secure its abstention, in order to get policy through. Since its support is essential it can bargain to have its own, electorally endorsed policy accepted by the government and, if not, defeat the alternative proposed.

Why this is so is shown in Figure 1.1, where actors prefer any policy closer to their own preference on a Left‐Right continuum. This puts C, at the median, in the most powerful position. Policy‐motivated actors both to left and right need C to form a majority. C can thus bargain for a public policy close to its own position, by threatening to join the alternative majority if C does not get its way. Compared to the policy positions of its rivals on one wing, C's position will be preferred by parties on the other wing in whatever coalition it joins. Thus C's position will constitute the point to which majority‐backed

Choosing Governments or Identifying Preferences? The Role of Elections in Democracy

FIGURE 1.1. The policy dominance of the median actor C in election, committee, or legislative voting

(p.7) policy always tends, where voting is determined solely by the desire to advance one's own policy‐preferences.

Of course the matching of median party with median elector will occur only under a voting system that faithfully reflects popular votes in party proportions of legislative seats. Where the relationship is distorted the real median party may be excluded from median status in the legislature so the carrier of the popular preference is not given a policy‐veto. What happens under these circumstances can be illustrated by the case of Britain in the post‐war period.

Elections are a defective democratic mechanism in Britain because the popular majority are only occasionally allowed a pivotal position in the legislature. The single‐member district (SMD) system operating in Britain generally awards a plurality party, which will have won 37–48 per cent of the national vote, the majority of parliamentary seats. It then forms a government that can do whatever it wants under the unwritten constitution, subject only to encountering extra‐parliamentary resistance (Budge et al. 1998: 177–98, 681–700).

The situation may not be as anti‐democratic or anti‐majoritarian as it seems where the legislative majority and the governing party that it supports is also the median party judged in Left‐Right terms. Table 1.1 shows the extent to which this has been so, during the post‐war period.

Over the course of sixteen elections from 1945 to 2001 inclusive, the median party, even very broadly conceived, became the legislative majority

TABLE 1.1. Equivalence between the median electoral party and the government in Britain, 1945–2000

Election year

Post‐election government party

Middle party Left‐Right issues

Did middle party win?

1945

Labour

Liberals

No

1950

Labour

Conservative

No

1951

Conservative

Conservative (almost)

Yes

1955

Conservative

Conservative

Yes

1959

Conservative

Conservative

Yes

1964

Labour

Labour (almost)

Yes

1966

Labour

Labour

Yes

1970

Conservative

Conservative

Yes

1974(i)

Labour

Liberals

No

1974(ii)

Labour

Liberals

No

1979

Conservative

Liberals

No

1983

Conservative

Liberals‐SD Alliance

No

1987

Conservative

Liberals‐SD Alliance

No

1992

Conservative

Liberal Democrat

No

1997

Labour

Labour

Yes

2001

Labour

Labour

Yes

(p.8) eight times, or after only half the elections. What is interesting and significant is that six out of the eight occasions when the middle party, broadly defined, got into office occurred between 1951 and 1970! In 1974 a relatively extreme left‐wing Labour Party won a parliamentary majority, then an increasingly right‐wing Conservative party got into government from 1979 to 1997. In other words, there has been a tendency for parties taking an extreme ideological position rather than one based on the ‘middle ground’ to win majorities from 1971 onwards. In that sense British governments have moved further away from reflecting majority opinion in the recent post‐war period. Only ‘New Labour’, by shifting substantially rightwards, managed to reoccupy the centre in 1997 and 2001 and make itself a more consensual government.

The consequence of the mismatch between moderate majority preferences and extremist government policies from 1974 and 1997 is agreed by most commentators to have been a widespread loss of support for political institutions in general, popular cynicism about the real extent of democratic choice and unprecedented political apathy by 2001 when less than 60 per cent of electors actually voted in the general election of that year. Rule by the plurality party is justified in terms of ‘working’ mandate theories when they have to cope with the absence of a real majority party (cf. Powell 2000), that is, in all elections except for a few presidential ones in the USA. In this situation the party nearest a popular majority is justified by such theories in taking over government and putting its programme into effect unimpeded. Such a party may, however, like the Thatcherite Conservatives, be strongly opposed by the actual majority. Certainly the British electorate seems by its reactions to endorse the alternative view that compromises produced by empowering the median position are more acceptable. Regardless of who forms the government, the party ‘carrying’ the median preference of the electorate should have a pivotal voting position in the legislative. Even if it just follows its own policy preferences rather than taking on a more general representative role, the coincidence ensures that public policy corresponds as closely as possible to the expressed preferences of all citizens, and certainly to the majority of them. Few would characterize the workings of the ‘elective dictatorship’ in Britain over the last thirty years as being very democratic.

1.2 HOW ELECTIONS WORK FOR A DEMOCRACY

Democracy should entail popular specification of public policies. This is clear from a normative point of view (Saward's ‘necessary correspondence’ of policy and felt interests, Saward 1998: 51) and as a research conclusion from our own investigation later. In taking this position we draw on the oldest conceptions of democracy. Voting in Greek city states was held directly on policy proposals and governments were not elected at all but drawn by lot (Bonner 1967). This view is also of course the driving force in (p.9) modern support for direct democracy, that is, popular voting on policies (Budge 1996). The specification of governments was only conflated with specification of policy in the eighteenth and nineteenth centuries, as a way of extending elective practices to large populations and territories. Electors in a large state could not decide directly on policy, as they could not meet together to discuss it (so it was argued). But they could elect representatives to discuss policies for them (Mill 1861/1910: 179–80).

With the emergence of political parties, electors could choose between or among policy stances as indicated by the parties' election programmes. The usual take on elections, however, places much more attention on which parties would form governments rather than policy adoption and implementation. For one thing, depending on the party system, forming governments often called for post‐election party negotiations. For another, even single‐party governments face many obstacles to implementation. The last two complications can obscure party responsibility and accountability for what had actually been done by the government. Policy is a parliamentary and governmental matter to be studied after the dust of the election has settled and the winners determined.

Viewed purely as mechanisms for selecting governments, elections appear as remarkably clumsy and approximate instruments, to the extent that most analysts and commentators are resigned to popular control over public actions not really being exercised (Powell 2000). SMD systems provide a strong incentive for two dominant political parties to contest elections within a district (Duverger 1954; Katz 1980; Riker 1982). The evidence on parliaments elected through SMD systems indicates that the mechanics of an electoral system can translate votes into seats in an odd way. This is a consequence of SMD's tendency to translate the vote percentage of the leading party into a seat percentage that is much larger. In this way, the SMD system frequently manufactures parliamentary majorities out of electoral pluralities. That alone is troublesome inasmuch as less than a majority of the people have given their consent—or a majority may actually be opposed (cf. the British case cited above). Even more troublesome for notions of popular control is the fact that certain geographical distributions of votes allow a second‐ranking party in SMD systems to have its votes translated into a majority of the seats.

The manufactured majorities and other oddities of the vote‐to‐seat translation under SMD systems give force to arguments for using proportional representation (PR) to elect parliaments. Arguments in favour of a close party seat to party vote linkage in the legislature are very forceful. However, it is far from settled as to whether PR systems do provide a reliable connection between electorates and governments. The reality for parliaments elected through PR rules is that usually no party wins an outright majority of seats, and multiparty governments form on the basis of post‐election bargaining, which may or may not square with the election results. Downs (p.10) (1957: 142–63) reasoned that voters in multiparty systems are not able to predict the party coalitions that form governments after an election. It is thus difficult for voters in PR systems to use their votes rationally, where the purpose is to put the reins of power in the hands of a party or parties that will do what the rational voter wants.

Theorizing and analyses of government formation in multiparty systems lend a good deal of credence to Downs' argument. Government formation in these circumstances is viewed in scholarly work principally as a matter of legislative consultation and negotiation, not much tied to election results. To the extent that election results do enter the theory and analysis of government formation in multiparty systems, it is mostly as backdrop. Parties use seats as a resource when trying to figure out what is in their best interest during the negotiations. Riker (1982) and DeSwaan (1973), for instance, suggest that one party might coalesce with another in a policy‐blind calculation based on the spoils they can gain by forming a minimal winning coalition. A minimal winning coalition could be conditional on policy preferences, in the sense of Axelrod's conjecture (1970) that government coalitions are policy connected, that is, bring together ideologically adjacent parties. Even with policy connectedness, however, a centre party in a predominantly three‐party system, such as the Dutch, could by itself and regardless of the election results choose whether it wants to push the policy position of government to the right or to the left.

All the evidence indicates that most governments form relatively autonomously of elections. So what role do elections play in democracy? Our suggestion is that they specify the median (majority) preference of the population, plus the overall policy structure or space within which it is embedded. Politicians and parties then operate within this revealed structure, which a properly functioning electoral system will reproduce in terms of party vote shares in the legislature. These will give a decisive voice on policy to the median party. Election results both empower the median position and inform politicians what it is, thus eliminating the possibility of strategic miscalculations messing up median‐(majority‐)based outcomes.

The electoral process therefore, if it functions properly in democratic terms, creates the ‘necessary correspondence’ required by normative democratic theory ‘between acts of governance and the equally weighted felt interests of citizens with respect to these acts’ (Saward 1998: 51). It does this in both a cognitive and empowerment sense. But in many ways the informative and communicative element comes first. This is because politicians would not even be able to react in strategic or power terms if they did not know the shape of the relevant policy configuration. They have to share a sense of what the underlying policy dimension is and how parties and electors are arranged on it, in order to decide on appropriate action and to form alliances. If they do not know what party is at the median for example, or if (p.11) they all had differing perceptions of it, their resulting behaviour would be wildly erratic and certainly stand little chance of systematically translating electors' preferences into policy.

All this of course implies that there is a median position to be taken and an ordered structure of preferences underpinning it. Mathematical analyses have cast doubt on a median necessarily existing in Euclidean spaces of three or more policy dimensions (McKelvey 1979) and even with two policy dimensions (Schofield 1985). In terms of pure issue spaces a median position can only be absolutely guaranteed in one dimension and by extension, in policy‐spaces with separable dimensions (Ordeshook 1986: 250) or with correlated dimensions (Adams and Adams 2000). This is already much, since actual issue spaces are usually of these types. Chapter 2 shows that the real‐life world of representative democracy, where a limited number of political parties bind together policies into packages and simplify choices for electors, largely evades all these theoretical difficulties and to all intents and purposes guarantees the emergence of a median policy position.

The argument for elections having a policy specifying rather than a government specifying function does not depend therefore on the existence of a one dimensional election space. However it is facilitated by it. We argue in Chapter 3 that spaces of this type emerge from the Left‐Right, bipolar, terms in which electoral debate is usually conducted. The election campaign itself is an active force simplifying and compressing other issues into such a space. The dynamics of political rhetoric and media simplification for a mass audience require straightforward summary comparisons to be made between national parties. In turn these squeeze out regional variants and group concerns, along with peripheral issues, by focusing on their central confrontation, which is interpreted in terms of a single Left‐Centre‐Right continuum.

Thus the median voter and his or her preferred party are identified by the election. Research reported below shows that this party tends also to be at the median on the majority of the separate issues into which the policy specialization of ministries and complementary division of legislative labour divide debate in the interelection period. This gives the party a general influence over public policies made by legislative processes, which in turn enables it to bring them closer to its own and hence to its median supporters' preferences and thus to those of the popular majority.

1.3 RE‐EVALUATING ELECTORAL PROCESSES

Not only is a switch from a government‐centred to a median‐centred mandate theory practicable, it also confers other theoretical and normative advantages. These in themselves make a compelling case for adopting an alternative, policy‐based view to the ‘choice of government’ one.

(p.12) Suspend the thought that elections are about which party governs and the world of mass democratic governance looks very different. Instead of asking who governs, ask from what policy position does governance emanate? The answer is simple in a uni‐dimensional policy space with political parties motivated by policy. The position in control is that of the median party in parliament, which Peter van Roozendahl identifies as the ‘central’ party (Van Roozendahl 1990; 1992). If there is a central party, which in a single dimensional space there almost always is but for the special case of the policy point falling between two parties, policy proposals favoured by the central party will be controlling.

What could this mean for the role of elections? It implies that their purpose is to communicate where the median voter stands. It means that voting is an expressive act with a guaranteed value: every vote counts not in deciding the winner but in identifying the median. It means that whatever value there is to majority rule, a majority will be in control. To the extent that their choice is not limited by sparse offerings from parties and otherwise remains undistorted by the electoral rules translating votes to seats, it means the median party will be at a position reasonably close to the median voter. It also means that as long as the dimensionality of election space and the positions of the parties along the dimensions are understood, the allocation of political responsibility is clear. It belongs to the policy position of the median party. Finally, it means that authorization to act for the popular majority takes place through empowerment of the median party, which after the elections is motivated by its own policy driven incentives to control public policy.

Many of these insights have a normative as well as empirical bearing on democracy, covering points traditionally made by mandate theory about the legitimation of some party policies compared to others and who has accountability for their implementation. The generalized formulation avoids the difficulties traditional mandate theory faces with regard to the legitimation and accountability of non‐majoritarian or multiparty governments. In order to change policy or to punish non‐fulfilment of promises, change the median position—a more practicable task than changing a coalition government in toto or even ejecting an entrenched plurality government.

This policy‐based mandate theory of elections also makes a decisive contribution to a longstanding debate in political science: which election system is best, PR or SMD? It is clear that the latter, with its manufactured majorities based on the plurality (or even second) party, offers no guarantee in the short‐term2 of empowering the median and often in fact discriminates heavily against it. PR is more reliable in this respect.

(p.13) The traditional argument for SMD is that it aids the emergence of strong—often single‐party—governments. Governments that ignore or flout the median preference are not normatively desirable however. They contribute to adversarial stances which can in the end lead to suboptimal policy mixes (Lijphart 1999). The alleged defect of PR systems, that they produce weak coalition governments, seems less serious when it is realized that policymaking emanates from the median rather than from the government position.

What does the policy‐orientated view tell us is not important about elections? For a party to seek more votes by moving around the policy space is essentially futile with respect to disrupting the communication purpose of the election. If a party moves to the position of the median voter, and if voters believe the party movement is sincere, and if the party follows through, then no harm is done. The subsequent policy adoptions will fall in the vicinity of the median voter. On the other hand, if the party movement is just a strategic electoral manoeuvre without any follow through once in office and if voters fail to foresee that insincerity, the safety valve is that the other parties in parliament will line up along the dimension and oppose the policy proposals of the strategic party. That is, the true median party will re‐present itself in parliament and take control from a party that achieved power on a purely insincere basis.

A policy‐based view implies that incumbents should not necessarily be the targets of credit or blame for good and bad policy. What matters is the location of the median party particularly if it is a reasonable size. Taking a policy‐based view also implies that Downs' worry (1957: 120–45) about identifying future governments under PR rules is not terribly relevant. A voter's interest will be served only by casting a sincere vote for an expressive purpose.

A final advantage of median‐mandate theory is in toning down the sharp contrasts often drawn between direct and representative democracy. Direct democracy, though beset by many possible problems, clearly involves citizens in deciding public policy through devices such as referendums or most notably the initiative.3 Hence it meets Saward's criterion of a ‘necessary’ connection between popular preferences and public policy better than the classic conception of representative democracy, where policy depends in the end on the will of legislators rather than on that of the people. Indeed, the attraction of direct democracy for many is simply that it seems more democratic in terms of popular control—giving a powerful impetus to the growing use of referendums in the modern world (Mendelson and Parkin 2001; Auer and Bützer 2001).

(p.14) The contrast seems overdrawn however when we bring in political parties and their role in presenting policy programmes to the public in elections, which are then taken as deciding between the competing packages. Given the central role of the parties in contemporary democracies they have really become ‘party democracies’ where the parties stand in elections on the basis of their programme—a programme to which governments formed by them and legislators attached to them are bound by convention and long‐term considerations (Downs 1957: 95–104) to support during the interelection period.

All forms of mandate theory take this view of the way democracies function. The policy‐oriented view underlying the median mandate stresses it even more than the others, however, since effecting the preferred policy does not depend on a majority actually emerging to support a single party government. Electoral endorsement of policy comes through clearly under all circumstances if it takes the form of a median mandate.

Even with this, a party democracy offering electors a choice between bundles of policies endorsed by each party, might still produce different results from voting on each policy individually. Table 1.2 provides a simple illustration. A party might gain majority approval on its programmes even though a majority of electors would vote against its position on each single issue. There is no simple way of saying which would be a ‘correct’ result over the range of policies in the table. It may be that certain types of issues, notably Left‐Right ones, are naturally tied together in people's perceptions and hence best voted on en bloc, as a general orientation for policy. Issues outside this bundle, which have not as a result been really voted on in an election campaign, might be best decided by direct vote in a referendum.

TABLE 1.2. A possible voting situation in which party B wins on its overall programme even though a majority opposes its position on each specific issue

Issues Voters

X

Y

X

1

a

b

b

2

b

a

b

3

b

b

a

4

a

a

a

5

a

a

a

(p.15) Resolving this question however is not really our concern here. is the way in which a policy‐orientated view of modern democracies shifts the argument away from a simplistic confrontation of the merits of representative deliberation versus direct popular voting. What it shows is that both involve policy choices between the positions endorsed by parties (Budge 1996). But one may be a wider, more general choice than the other. Closing the gap with direct democracy is just one advantage of adopting a median mandate approach to representative democracy, to add to the others listed above.

1.4 EVALUATING THE MEDIAN MANDATE

The ability of median mandate ideas to meet many of the conceptual and practical problems that confront mandate theory in general make it hard to argue that a change in perspective is not worthwhile. In this life, however, the advantages of such a move are often balanced, if not outweighed, by the disadvantages. What are they in this case?

A first springs immediately to mind. The theory has something of a naïve, idealistic cast even if it also possesses some historical and normative force. It would indeed be nice if majority feeling were automatically reflected in policy processes whether or not a majority actually formed. But is this not too good to be true? Can parties be trusted to run with the preferred policy position? Will governments, with all their powers, really substitute the median position for their own (leaving aside the rare cases where they coincide as a single‐party (electoral) majority government)?

The belief that elections are about winning office is so common that a first response to questions about democracy is that it makes governments dependent on winning elections. There is, however, even among those who appear to accept this truism, a large body of commentary on how undesirable it is. Madison himself provides one example. Perhaps his most clever and powerful argument was to claim that overlaying a democratic apparatus on a large territory and population was a good idea because from time to time, issue to issue, there would be cross‐cutting cleavages and shifting coalitions so that there would be no permanent majority, that is, no permanent winner. In a contemporary commentary on Madison's idea, Guinier (1994) refers to the spirit of what Madison had in mind as ‘governing by taking turns’. Before Guinier, Dahl (1956) referred to Madison's idea as democratic pluralism.

Buchanan and Tullock (1962) offered The Calculus of Consent as a theory of democracy. In it, they take as their premise that the unanimity principle is the ethically endowed democratic decision rule, majority rule a mere mechanism for efficient decision‐making. As we have pointed out the median position is the best outcome everyone can hope for and hence can agree on. In a thoughtful and sobering commentary on the drift in thinking about (p.16) modern democracy in the USA, Cain (1992: 273–5) warns of a new populism that puts so much stock into the idea of winning, clearly and decisively, that it threatens the idea of representative government. To the criticism of naiveté we would say that all of these commentaries are telling us that, in one way or another, putting exclusive power in the hands of winners is not the desired outcome. Therefore, it is not naïve to want elections to operate in policy terms. Can they?

Clearly even the most desirable normative outcomes have to face the test of reality. If the conditions for their application do not exist in working democracies, normative hopes cannot justify them. On issues of practicality, one could quibble with several of the conditions that would have to hold in order for the world to work according to our ideas. The diversity of policy positions offered by parties may certainly be too sparse. Just as surely, the electoral rules can distort the vote to seat translation thus separating the median position in parliament from the median among voters. And—a possibility already considered—policy disputes may not naturally line‐up in the required policy space. Even if they did, all but the median party would have a strong policy incentive to rearrange to their own advantage that common way of viewing the world (Riker 1982). Finally, striving to achieve a preferred policy outcome is a dubious goal to specify for politicians. Even if the median party acts in its own self‐interest by pursuing its own preferred policies, will it not be forced into compromises or outright abandonment of them, possibly in exchange for office?

Granting that all these criticisms have some validity in the abstract, we should nevertheless consider the empirical evidence before engaging in extensive arguments over their plausibility. That is what we are going to do in the following chapters of this book, with specially collected information from functioning post‐war democracies.

1.5 PLAN OF DISCUSSION

Before doing so, however, we need to systematize and detail the general arguments rehearsed in this chapter. Here we have given the broad view. Now we need the detail, starting with the assumptions of mandate theories themselves, in Chapter 2. Not only do we summarize them in propositional form, we also make a preliminary check on whether some of their conditions are actually met in contemporary democracies.

Finding a median depends on the kind of issue space political debate and party competition is located in. A common tendency is to talk about it in Left‐Right terms. Is this justified? And what are the other possibilities? Median positions have to be recognized in order to exert any effect. How far do parties and electors communicate preferences to each other within a shared issue space, rather than one fragmented by steep boundaries of (p.17) knowledge and information? These questions are all considered in Chapter 3. Chapter 4 then focuses all these discussions into specific questions we can put to the evidence we have collected on established post‐war democracies in the second half of the twentieth century.

It is not of course as if mandate theories are totally descriptive, to be rejected if actual democracies do not operate according to their rules. On the contrary, both the government and the median mandate share the usual normative‐descriptive mix of political theory. Even if we find that contemporary democracies do not implement a government mandate for example this is not conclusive evidence that they should not do so. The mandate could still exist as a state to which they should aspire, because in some sense it might seem more democratic than what they have actually got.

On the other hand, since mandate statements also have a descriptive aspect empirical evidence does have some impact on the standing of the theory. This is especially true if an alternative, like the median, with equally good democratic credentials, does appear to match better with what these well‐accredited democracies are doing. One can at least say that the median provides a better description of the way they work and hence seems easier to apply in practice. To this extent, an empirically supported mandate theory can be said to constitute a relevant functioning norm for democracy.

These considerations underlie the checks undertaken with comparative and overtime evidence in Parts II, III, and IV of the book. Parties are the crucial catalyst of electoral policy choices in modern representative democracies. In Chapter 5 we consider how good parties are in this role, particularly how far they offer ‘a choice not an echo’. Because the government mandate fails all too often with respect to some of its own required conditions, an alternative or supplementary version has gained theoretical force. That is the idea that, if electors cannot reliably pick party governments on the basis of their future programmes, they can at least evaluate them on their record and reward or punish them accordingly. Chapter 6 reviews the electorate's ability to systematically hold governments to account in the context of ‘economic voting’, but concludes there is little evidence for it. On the other hand, Chapter 6 shows strong evidence that median voter positions could be empowered by translating them accurately into similar policy positions of governments. Chapter 7 investigates how, despite representational distortions deriving from limitations on party policy offerings, electoral system mistranslations, and post‐election party negotiations, median voter and government policy positions are generally aligned. The key finding is that while distortions are readily apparent everywhere, across the steps of the representational process and more especially through time, the distortions tend to cancel one another out so that in the intermediate and long run distortions do not amount to much in terms of representational bias.

(p.18) Part III examines the correspondence between median voter preferences and actual public policy. It begins in Chapter 8 with an analysis of the connections between parliamentary medians, governments, and ministries on the one hand, and on the other the policies that governments declare their intention to pursue while in office. Chapter 9 takes the next step and looks beyond government policy intentions to investigate how preferences of the various actors connect to actual policies. There is only weak evidence that declared policy intentions and actual policies follow in any sort of predictable way from the Left‐Right or other policy preferences of median parties or median voters.

Those weak results force us to step back from the complexity of everyday politics and policy and reconsider the policymaking process in a larger democratic context. Having found in Chapter 7 that representation is distorted in numerous ways in the short run but is accurate in the long run, we look at the short and long run dynamics of policy and politics in Chapters 10 and 11. Our investigations show policy has a slow‐moving dynamic while electoral politics and their impact on parliaments and governments are highly dynamic. We also find in virtually every one of our twenty‐one democracies that highly dynamic politics are anchored in long‐term equilibrium (Left‐Right) positions. In Chapter 12, we investigate whether these long‐term equilibria are responsible for setting the mark for slow moving policy, what we label the policy regimes, in our twenty‐one countries. In three different policy areas—the size of a nation's political economy, the resources it expends on its welfare state, and its peaceful versus militarist orientation to international affairs—we find this to be the case. Further, the political preferences with the strongest and most consistent policy effects are those of the Left‐Right parliamentary median.

We conclude with Chapter 13, showing how a median mandate theory pulls together strands of democratic theorizing that until now have stood as qualifications on democratic possibilities. Our account provides a rounded view of representative democracy centred on the median mandate. With that discussion we come full circle, back to the central propositions of mandate theory, which we now present and discuss in Chapter 2.

Notes:

(1) Note, however, that the same arguments would apply to a situation where votes were weighted by the intensity of feeling behind them. It is just that the location of the median would differ in this case.

(2) Taking a long‐term view, the alternation of plurality‐based governments under SMD, even where they are generally to the Left or Right of the electoral median, does produce an average policy effect that can approximate the average median elector position. In this sense SMD systems are more representative than they seem when one takes them government by government (cf. Chapter 12).

(3) Both referendums and initiatives involve electors voting for and against policy proposals. The former however are often called by governments at their own convenience while the latter can be called independently by the proposers of some policy change, thus putting the calling of a vote as well as participation in it into the popular domain.