tiatives are supposed to outweigh the benefits of additional control. Moreover, direct democratic de-
cision-making is said not to be useful for decisions at the supranational level of a confederation be-
cause its potential for compromise is very much restrained. Binding compromises are however im-
portant in international organisations for mutual agreement of states on proposals in their mutual inter-
ests. Referenda and initiatives would then not allow for decisions in the long-term interest of a world
order because citizens focus on their national interests.
The small scale argument is not completely convincing because the benefits at stake increase at least
as much as the costs if direct democratic decision-making is introduced at higher levels of govern-
ment. The agency problem is the more severe the higher the level of decision-making is in a federa-
tion. Potential deviations from citizens’ interests lead to higher expected costs for these citizens at the
national than at the local level because the importance of the political issues increases. Moreover,
many countries, among them several European ones, also have a history of referenda for important
constitutional changes. It is for example meanwhile common in the U.K. to have a referendum on the
participation in additional steps of political integration in Europe. The legitimacy of such referenda is
not challenged at all on grounds of a bigger scale polity. The higher organisational costs of conducting
referenda and initiatives at the EU level as compared to the local level is also not convincing. Europe-
wide referenda may pose additional organisational difficulties as compared to national referenda
which could however be coped with in modern information societies. The confederation argument
does not hold because the EU Treaty already has considerable elements of a federation. The more
elements of a federation a polity has, the easier arguments for a direct participation of citizens in po-
litical decision-making can be made. From the perspective of optimal control of representatives, it is
even necessary to introduce referenda and initiatives as control instruments as soon as the process of
federation building starts.
There is a second asymmetric information case if representatives are imperfectly informed about citi-
zens’ preferences. Matsusaka (1992) argues that under such circumstances even benevolent politi-
cians may impose policies that are deviating from citizens’ wishes. Hence, they have incentives to
propose a referendum whenever they are uncertain of citizens’ wants in order to avoid being pun-
ished at the polls for enacting the wrong policy. The author provides evidence that referenda in Cali-
fornia rather are on distributive issues than efficiency/ procedural issues. Referenda are also used in
periods of political corruption to restrict government. This argument has implications for the EU. The
further representatives are away from citizens, the less informed they are about citizens’ political
preferences, which holds more strongly in the EU than in the local case. Indeed, the EU is more
strongly engaged in redistributive politics such that it pays off for EU representatives to elicit citizens’
preferences via referenda.
The second argument of opponents against direct democracy is that imperfect and asymmetric infor-
mation may provide opportunities for interest groups to unduly influence political outcomes in direct
democracies. Matsusaka and McCarty (2001) analyse the case when interest groups are able to fool
representatives. They use the threat of an initiative by pretending that their position is closer to citi-
zens’ preferences than the proposal of representatives, although this moves the policy outcome fur-
ther away from the ideal point of the median voter. According to Gerber (1999), interest groups
certainly influence outcomes in direct democracies, but they also have an impact on policy outcomes