The name is absent

include both single and coalition in the same regression, but we use them as al-
ternatives to assess the robustness of the results. (The two measures correspond
to alternative definitions of the default type, minority governments being lumped
together with single-party governments when coalition is used, and together with
coalition governments when single is used.) When we take the average of these
two indicator variables (in the cross sectional estimates or in the course of a legis-
lature), we thus obtain two incidence measures: coalition measures the incidence
of coalition governments over the relevant time period, and likewise for single. As
an alternative measure of government type, we use the number of parties in the
governing coalition, ngov.

Other constitutional variables To identify some of the causal effects de-
scribed in the theory, we need to measure some other features of the constitution
likely to influence either the party structure or the type of government. In par-
ticular, in some electoral systems a party has to overcome a minimum electoral
threshold to gain representation in the legislature. We call this variable threshold,
and measure it as a percent of the total vote at the national level. A higher thresh-
old is expected to reduce party-fragmentation. We also construct (0,1) indicator
variables for three additional constitutional features: investiture, set to 1 if a new
government must win majority support in an investiture vote; constructive, set
to 1 if the government can be dismissed only by electing a replacement; bicam, set
to 1 if the legislature has two chambers (irrespective of their relative strength).
Bicameralism raises the probability of surplus coalitions (since it increases the ma-
jority required to form a government or to pass legislation), and thus we expect
it to have a positive effect on the incidence of coalition governments, a negative
effect on that of single-party majority governments. The investiture vote and the
constructive vote of no confidence reduce the probability of minority governments,
and thus, for a given party structure, are expected to increase the incidence of
coalition governments, but have no expected effect on single-party majority gov-
ernments — see, in particular, the recent work by Diermeier, Eraslan, and Merlo
(2003a and b).²¹

Both political outcomes and government spending may be influenced by the
quality of democracy. We measure it by the Polity IV variable discussed above, in-
terpolated with the Freedom House data set when missing. The resulting variable
is called polity_gt.

²¹ The variable constructive should be treated with care, however, since only two countries
in our sample have this constitutional feature.

35

<<   33   34   35   36   37   38   39   >>

More intriguing information