- 12 -
the referendum or initiative decisions at the EU level helps to transform individual self-interest of citi-
zens to a public interest and the national to European public interests. The nationally informed posi-
tions of citizens may be questioned according to what extent they can be generalised to the European
level. In order to make such reflected political decisions at the EU level, citizens need incentives to
discuss European issues. Referenda and initiatives at the EU level are a convenient way to provide
them.
2.5 The Wicksellian Connection between Public Services and Tax Prices
The most convincing argument for the introduction of instruments of direct democracy at the EU level
stems from the fact that referenda and initiatives allow to conduct public policy such that the tax
prices citizens pay for public goods and services are linked to the benefits the citizens obtain from
these public goods and services. Breton (1996) calls this the Wicksellian (1896) connection. The
systematic empirical analyses of the impact of referenda and initiatives on economic policy for Swit-
zerland and the U.S. states provide strong support for the hypothesis that the Wicksellian connection
rather exists in direct than in representative democracy.14 According to these studies, public spend-
ing,15 revenue16 and debt17 are significantly lower in jurisdictions with direct democracy than in those
with pure representative democracy. With respect to spending structure, the fiscal referendum mainly
restricts welfare and administrative spending according to Schaltegger (2001) and Vatter and Freitag
(2002). With respect to revenue structure both, the U.S. states with initiatives and the Swiss cantons
with a fiscal referendum, rely more on user charges than on broad-based taxes.18 Moreover, Ma-
tsusaka (1995, 2002) and Schaltegger and Feld (2001) provide evidence that centralisation of
spending and revenue is reduced by the referendum.
The question remains, however, whether the lower level of public spending also leads to a more effi-
cient public sector. Pommerehne (1983) analysed costs and prices of local garbage collection in 103
Swiss cities in 1970. He found that average refuse collection costs (per household) were - ceteris
paribus - lowest in cities with direct legislation and private garbage collection. For a panel of the
Swiss cantons from 1970 to 1996, Barankay (2002) reports significantly lower infant mortality rates
and a higher share of college degrees in more direct democratic cantons suggesting that this indicates
a higher quality of public goods in the cantons. Pommerehne and Weck-Hannemann (1996) show
that in those Swiss cantons in which citizens have an impact on budgetary policy in direct legislation,
tax evasion is - ceteris paribus - lower as compared to the average of the cantons without such di-
rect influence.19 These results are corroborated by Feld and Frey (2002a) and by Torgler (2002)
14. For earlier descriptions of these studies, see Kirchgassner, Feld and Savioz (1999, chap. 5), Feld and
Kirchgassner (2000, 2001) and Matsusaka (2002).
15. See Matsusaka (1995, 2000, 2002) for the U.S. and Feld and Kirchgassner (1999, 2001), Feld and Matsusaka
(2003), Schaltegger (2001) and Vatter and Freitag (2002) for Switzerland.
16. See Matsusaka (1995, 2000, 2002) for the U.S. and Feld and Kirchgassner (2001) for Switzerland.
17. See Kiewiet and Szakaly (1996) for the U.S. and Feld and Kirchgassner (1999, 2001, 2001a) for Switzerland.
18. See Matsusaka (1995) for the states and Feld and Matsusaka (2003a) for the Swiss cantons.
19. There are also theoretical arguments why citizens in direct democracies evade taxes to a lesser extent than
those in representative democracies. See Pommerehne, Hart and Feld (1997) and Feld and Frey (2002).