was at about the Swiss level: In the seven presidential elections between 1972 and 1996 it was 48.1
percent on average and 51.3 percent in 2000. In midterm elections, it amounted to 35.5 percent with
a minimum of 33.1 percent in 1990.11 These turnouts cannot be explained by the rational actor ap-
proach as long as it takes only instrumental voting into account.
Brennan and Lomasky (1993) suggest to resolve this paradox by a theory of expressive voting. Ac-
cordingly, voters use elections to express their discontent with specific policies of a government, but
also to show solidarity with particular policy platforms or to demonstrate a specific position, ideology
or habit. Instead of being instrumental in the sense of an action to achieve a certain end as in the case
of normal market transactions, voting would be expressive like people’s cheering at a football match
(Brennan and Hamlin, 2000, p. 130). If voting is expressive, there is a considerable concern that citi-
zens vote irresponsibly by expressing their political preferences for or against general or particular
policies because their voting decision may result from particular enthusiasms and prejudices. Brennan
and Hamlin (2000, pp. 176) thus question that direct democracy will lead to an orientation of politi-
cal decision-making at private or public interests. In contrast, they argue that pure representative
democracy is a better procedure in a world of expressive voting because it allows to reduce the ex-
tent of irresponsible political outcomes by a selection of representatives. Voters are supposed to
have a tendency to support candidates with relatively higher civic virtue because they cannot avoid
forming and expressing opinions on candidates’ competence and general political attractiveness.
These opinions may be informed by moral codes individually obtained during socialisation of citizens.
Reconstructing a basically Madisonian argument, representatives would hence be more public spir-
ited, conscientious and competent than those whom they represent (Brennan and Hamlin, 2000, p.
180). As a result, people tend to be better judges of other people than they are of competing policy
options.
The discussion of direct democracy by Brennan and Lomasky (1993) and Brennan and Hamlin
(2000) is flawed for several reasons. First, the authors fully neglect the discussion process that pre-
cedes direct democratic decisions, but also elections. From their arguments, it appears as if citizens
decide at the ballots after they have reflected on different proposals in isolation. In fact, decisions in
referenda and initiatives take place in a totally different environment. Frey and Kirchgassner (1993),
Kirchgassner, Feld and Savioz (1999) as well as Feld and Kirchgassner (2000) analyse the delib-
eration in direct democracy and characterise it as a relatively rational process which is the more in-
tense, the more important a political decision is, and which is informed by the different positions of
political parties, interest groups, but also by experts and less organised individuals. In the course of
this discussion, a learning process occurs at least in parts of the citizenry. Since many citizens are
confronted with the arguments of both sides, those opposing and those favouring a certain policy
outcome, they are induced to consider each proposal anew. This can lead to a revision of their indi-
vidual position. Therefore, the possibility also emerges that citizens examine the extent to which their
preferences generalise. In addition, it can be presumed that citizens’ willingness to bear costs of in-
formation acquisition is higher in direct legislation than in representative democracies. The reason for
this is that it can become privately important for citizens to be well informed about political issues.
Such a situation emerges if other citizens belonging to one’s social network expect a citizen to be
11. See: Statistical Abstract of the United States, 1997, p. 289.