Professional Documents
Culture Documents
1.1 Introduction
This volume addresses the contrast between the view that elections are a
mechanism to hold government accountable and the view that they are a
means to ensure that citizens’ views and interests are properly represented in
the democratic process. It intends to explore how far this contrast and its
embodiment in institutional structures influence vote choice, political par-
ticipation, and satisfaction with the functioning of the system of political
representation and democracy in general. The volume is mainly based on
data from the second module of the Comparative Study of Electoral Systems
(CSES) which had the same purpose.1 In this introduction we will further
expand on these different models of democracy and summarize how the fol-
lowing chapters contribute to our knowledge of the effects of different insti-
tutional arrangements on people’s political behaviour and their evaluation of
the functioning of democracy.
Ever since Schumpeter defined ‘the democratic method’ as ‘that institu-
tional arrangement for arriving at political decisions in which individuals
acquire the power to decide by means of a competitive struggle for the peo-
ple’s vote’ (Schumpeter 1976, 1942), there is a broad consensus that competi-
tive elections are the identifying property of the contemporary democratic
process (Powell 2000).
However, as much consensus there is on the importance of competitive
elections for democracy, so little agreement there is on the precise function of
elections. Different theories of political representation have different views
on this. Most modern conceptions of democracy agree that the basic function
of elections is to connect the policy preferences of the people to public policy.
1
Jacques Thomassen
However, they differ in what is meant by ‘the people’ and in the mechanism
that is supposed to connect the policy preferences of the people to govern-
ment policy. The two main views of democracy are the majoritarian and the
consensus or proportional view of democracy (Lijphart 1984; Lijphart 1999;
Huber and Powell 1994; Powell 2000). According to Lijphart (1984: 4) the
main characteristic of the majoritarian view is that when the people are in
disagreement and have divergent preferences, the government should pri-
marily be responsive to the majority of the people. In the consensus view
government policy should be responsive to as many people as possible. These
different principles lead to different views on the function of elections and
are embodied in different political institutions.
In this volume we focus on the distinction between the majoritarian and
consensus models of democracy. These models are the embodiment in insti-
tutional structures of the two different views of democracy and reflect a dif-
ferent view on the functions of elections.2 In the majoritarian view the single
most important function of an election is the selection of a government. It
requires that the voters have a clear choice between two competing (groups
of) parties. The concentration of power in the hands of an elected majority
government brings the government under tight control of the majority of the
electorate. This control can be based on two different mechanisms, depend-
ing on voters’ time perspective or rather the considerations they take into
account when they decide how to vote. If voters base their choice on what the
(two) competing parties have on offer in their party manifestos, the winning
party can be said to have a policy mandate from a majority of the electorate
(Powell 2000: 8). This is basically the mechanism assumed by the Responsible
Party Model (Schattschneider 1942; APSA 1950; Thomassen 1994; Klinge-
mann et al. 1994).
However, this model of political representation is very demanding, in par-
ticular with regard to what is required of the voters. Also, a single vote can
hardly provide a policy mandate for a multiple package of issue dimensions.
Therefore, this model is often claimed to be totally unrealistic and unfeasible
(Riker 1982; Thomassen 1994). An alternative and perhaps more feasible
model is based on Schumpeter’s idea of a competitive democracy. In this
model elections are an accountability mechanism, where the sanctions are to
extend or not to extend the government’s tenure (Schumpeter 1976 (1942),
Chapter 22).
The major difference with the policy mandate model is that voters make
their vote choice on the basis of their evaluation of the performance of the
incumbent government. If they are satisfied with that performance, they will
vote for the party or parties in government; if they are dissatisfied, they will
‘throw the rascals out’. This model of accountability is far less demanding of
the voters because all they need to know is which party, or coalition of
2
Representation and Accountability
parties, is in power and which one is the opposition. Their information about
the content of government policy can be limited. Being satisfied or dissatis-
fied with the government, its policies, or the outcomes thereof is all it takes
(Fiorina 1981).
An essential requirement of this model of accountability at the system level
is the clarity of responsibility (Manin et al. 1999; Powell 2000; Lundell 2011).
Accountability is by definition close to impossible if it is not perfectly clear
who, i.e. which political party or coalition of parties, is responsible for gov-
ernment policy. But not only the incumbent but also the possible alternative
future government must be identifiable. A second requirement is that the
voters’ sanction of the party or parties in power is effective, i.e. that they
really can throw the rascals out without the risk that these (or some of them)
will return to power after having lost the elections. This mechanism can only
work in a majoritarian system where two (blocks of) parties compete for a
majority of the votes and the winner automatically takes (over) government
responsibility. A final requirement is that the voters do indeed hold the gov-
erning party or coalition of parties accountable for their performance while
in office, i.e. when they vote retrospectively.
In consensus models of democracy, or proportional systems, the major
function of elections is to elect the members of parliament who together
should be as representative as possible of the electorate as a whole. The crite-
rion for the democratic quality of the system is how representative parlia-
ment really is. There is no deterministic relationship between the election
outcome and the formation of the government. As a multiparty system is one
of the characteristics of a consensus model of democracy, usually a coalition
of several parties will be needed to form a majority government. Coalitions
will usually be broad, doing justice to the principle that not just a bare major-
ity but as many people as possible can influence government policy.
Between them, the two models of democracy fulfil the two most important
functions elections in a representative democracy have, according to main-
stream normative democratic theory. First, elections allow voters to deter-
mine the political colour of their government, making government
accountable to the judgement of the people. Second, elections produce a leg-
islature that is representative of the distribution of policy preferences amongst
the electorate. However, it may be obvious that there is a certain tension
between these two functions (Manin et al. 1999; Powell 2000; Lord and Pollak
2010; Aarts and Thomassen 2008a; Dalton et al. 2011a; Dalton et al. 2011b).
Electoral and more generally democratic systems cannot optimally serve both
functions at the same time. An important criterion of democracy underlying
the consensus model of democracy is inclusiveness, i.e. the part of the elector-
ate represented in government is as large as possible (Lundell 2011; Kaiser
et al. 2002). However, the better this criterion is met, the lower the clarity of
3
Jacques Thomassen
4
Representation and Accountability
5
Jacques Thomassen
A. Macro characteristics
of democratic systems
Majoritarian vs. consensual
Institutional quality
Policy performance
Clarity of responsibility
Polarization
(AC). Most of the research questions in this volume can only be answered by
taking into account one of these types of interaction.
6
Representation and Accountability
7
Jacques Thomassen
8
Representation and Accountability
9
Jacques Thomassen
10
Representation and Accountability
individuals becoming more prone to say that a party represents their views.
When perceptions of representation are taken into account the—modest—
relationship between PR and turnout loses its significance, i.e. proportional
representation has no direct, independent effect on electoral participation
once political attitudes are taken into account. In other words, the relation-
ship between PR and turnout is indeed mediated by individuals’ ability to
identify a party representing their views.
Chapter 7, by Weldon and Dalton, examines how democratic institutions
affect citizens’ political behaviour beyond the act of voting. The starting
point is again Lijphart’s presumption that consensual institutions incorpo-
rate more citizens into the electoral process and lessen political inequality,
because they give citizens effective voice and representation. This presump-
tion seems to imply that consensualism has spillover effects that stimulate
participation more broadly. However, in contrast to the effect of consensu-
alism on turnout, these further implications of consensualism have hardly
been examined and are not uncontested, not even at a theoretical level.
According to a competing view, consensualism may have the opposite
effect, actually decreasing citizen involvement beyond voting, precisely
because citizens already feel better represented through the electoral pro-
cess in these systems. In addition, majoritarian electoral systems tend to be
candidate-centred, with MPs elected from single-member districts. Because
of this, majoritarian systems offer greater incentives for participation
beyond voting, such as campaigning and contacting public officials. On the
second dimension of political institutions, the federal–unitary dimension,
federal systems may suppress voter turnout, because individual elections
are less critical, but they also offer more opportunities for participation at
the sub-national level than do unitary systems. Because elections occur
more often, civic organizations may stay stronger over time, helping to
mobilize and encourage non-voting forms of participation. These hypoth-
eses are corroborated by the empirical analysis. Participation in all other
political activities than voting is lower in consensual systems when opera-
tionalized on the basis of the executives–parties dimension—often mark-
edly so as in the case of campaign activity, persuading others how to vote,
and contacting a politician. In short, consensual systems seem to demobi-
lize citizen participation beyond casting a ballot. These correlations are
even more outspoken when only the electoral system is taken into account.
As shown in the previous chapter, a proportional electoral system might
encourage people how to vote, but it apparently discourages people from
showing their support during the campaign or trying to persuade others
how to vote. These patterns are unanticipated by the consensualism litera-
ture, which presumes that voting turnout is symptomatic of general involve-
ment in the political process.
11
Jacques Thomassen
12
Representation and Accountability
13
Jacques Thomassen
14
Representation and Accountability
citizens’ evaluations of regime procedures, i.e. their beliefs that human rights
are respected, their perceptions of corruption and the representation process,
are shaped by objective indicators of the quality of output institutions.
The alternative explanation that popular evaluations of regime procedures
are due to the ‘kinder, gentler’ consensus institutions that Lijphart envisioned
is hardly supported by the data. There is some evidence that citizens residing
in federal regimes are more likely to think that human rights are respected in
their country; and they also believe that corruption is lower. But all in all
there is fairly weak evidence in support of the idea that consensus systems
have more procedural integrity than majoritarian institutions.
15
Jacques Thomassen
and political leaders and therefore are becoming less satisfied with the func-
tioning of democracy.
The additional hypothesis tested in this chapter is that in consensus democ-
racies this effect will be subdued because one of the main characteristics of a
consensus democracy is that it is responsive to minority groups. Because of its
proportional electoral system, new political movements based on new con-
flict lines, such as globalization, have easy access to the process of political
representation.
The main hypotheses in this chapter are not borne out. Although in general
less-educated and less-skilled citizens do feel less well represented by any polit-
ical party and are less satisfied with the functioning of democracy, there is no
evidence for a differential development of these feelings: there is no growing
gap in this respect between the winners and losers of globalization. Since no
effect of globalization was found, this by definition means that the political-
institutional context is not relevant either.
Chapter 12, by Chang, Chu, and Wu, focuses on the well-known winner–
loser gap. Previous research has shown that among voters the losers of elec-
tions tend to be less satisfied with the functioning of democracy than the
winners. It has also been shown that this gap between winners and losers is
larger in majoritarian than in consensus democracies. In these previous stud-
ies electoral winners and losers are distinguished on the basis of a single elec-
tion and the interaction between winners and losers is treated as a one-shot
game. However, since elections in democracies are held routinely, winners
can become losers when electoral alternation occurs. In this chapter two con-
secutive elections are taken into account to define winners and losers. This
logically leads to four categories: two-time winners, two-time losers, winners-
losers, and losers-winners. In this chapter this distinction is connected to
Huntington’s two-turnover test in the comparative democratization litera-
ture. According to Huntington, a nascent democracy is considered consoli-
dated if it has experienced two peaceful electoral alternations. However, the
relevant literature on this subject consequently neglects the importance of
ordinary citizens’ democratic attitudes during electoral alternations in young
democracies. And yet, it is generally agreed that citizens’ attitudes towards
democracy are essential for the legitimacy and stability of democratic regimes,
in particular new democratic regimes. Therefore, this chapter, by examining
how citizens’ experience as winners or losers in successive elections influ-
ences their support for democracy is an important complement to the two-
turnover test literature. Because of their limited experience with democracy,
the expectation is that the gap between winners and losers will be larger in
developing than in established democracies.
The main hypothesis at the individual level in this chapter is that the
experience of having been a winner at least once is more likely to develop a
16
Representation and Accountability
17
Jacques Thomassen
18
Representation and Accountability
emocracies. Still, these findings are too preliminary to justify strong con-
d
clusions, the more so since we did not really (re)examine to what extent
the winner–loser gap is larger in majoritarian than in consensus democra-
cies. Nevertheless, by extending the definition of winners and losers to two
consecutive elections the final chapter seems to deliver strong support for
Lijphart’s initial argument that different societies ask for different political
institutions. If winners and losers do not regularly change places because
solid social cleavages always yield the same political majority, a majoritar-
ian system is disastrous for the legitimacy of the political system among
the (permanent) minority of the people (Lijphart 1984).
But the general conclusion of this volume is that formal political institu-
tions are less relevant for people’s attitudes and behaviour than often pre-
sumed. This is not to say that characteristics of the political system do not
matter. But rather than formal political institutions like the electoral system
it seems to be characteristics of the party system like polarization and the clar-
ity of responsibility that really matter. This is in line with the growing body
of empirical knowledge, mostly based on the Comparative Study of Electoral
Systems, on the relationship between aspects of the political system and the
attitudes and behaviour of individual citizens (Thomassen 2005b; Klinge-
mann 2009; Dalton et al. 2011a; Dalton and Anderson 2011).
This conclusion should be a lesson in modesty for institutional engineers.
If institutional differences have as little effect as this volume suggests, or if
their effects cancel out each other, institutional reforms based on the pre-
sumption that a change in institutions will improve the functioning of
democracy are doomed to lead to frustration among both the reformers and
the citizenry at large.
Notes
1. See <http://www.cses.org/>.
2. The following characterization of the two models of democracy is from Aarts and
Thomassen (2008a: 6–7).
3. This figure was first presented by Bernhard Wessels in 1998 at a CSES conference in
Berlin. Also, see Thomassen (2000) and Klingemann (2009).
19