land at the cantonal and municipal level.
6
This literature essentially compares
various policy outcomes in the economic realm like debt levels, GDP growth,
tax levels, etc., between entities allowing for referendums and entities not al-
lowing for referendums, while controlling for other factors. Many studies come
to the conclusion that referendums have positive effects. Government spending,
debt levels and taxes are lower, GDP growth is higher, etc. (Kirchg¨assner, Feld
and Savioz, 1999; Feld and Matsusaka, 2003; Matsusaka, 2004; Funk and Gath-
mann, 2006).
7
Some studies, however, fail to find an effect (e.g., Besley and
Case, 2003) or an effect which goes against the commonly held wisdom (e.g.,
Camobreco, 1998).
Beside the economic realm, some isolated studies have focused on a more
direct link between referendums and policies in other domains. Gerber (1996)
shows for laws on the death penalty and parental notification of teenage abor-
tions that in states of the US with referendums the adopted laws reflect more
closely the preferences of the states’ voters.
8
Similar results appear in Gerber
and Hug’s (1999) work on minority rights, while results questioning this general
conclusion appear in Lascher, Hagen and Rochlin (1996) and Camobreco (1998).
9
For Switzerland Vatter and R¨uefli (2003) show that the extent of referendums
affects health policies at the cantonal level in Switzerland, while Armingeon,
Bertozzi and Bonoli (2004) find similar effects for welfare policies. Fischer (2005)
finds effects of direct democracy on redistribution (see also Feld, Fischer and
Kirchg¨assner, 2003), on educational achievement, and crime rates, among other
policy areas.
Thus, at the subnational level quite strong evidence is present suggesting that
referendums affect policy outcomes. Very little, however, is known whether these
effects can also be found at the national level.
10
In addition, most of the empiri-
6
Kirchg¨assner, Feld and Savioz (1999), Lupia and Matsusaka (2004) and Matsusaka (2004)
provide excellent surveys of this literature.
7
Interesting to note is that Funk and Gathmann (2006), employing a different estimation
approach find a much reduced effect on government spending. In addition, according to their
result, the reduced spending at the cantonal level in Switzerland due to direct democracy is
compensated by increased spending at the local level in cantons with strong institutions for
referendums.
8
Gerber (1999) discusses these results in a broader context while Hug (2004) shows more
fine-grained institutional effects for these policies.
9
As Matsusaka (2001) and Hug (2001) demonstrate, however, these results are largely due
to a misspecified empirical model.
10
Kleinewefers (1997) criticizes authors generalizing their results from the subnational to the
4
Convention