Notes
1
Saiegh (2002) similarly integrates the level of executive-legislative relations and the
level of individual legislators in a two-stage model of executive-initiated and
congressionally-approved policy change (Saiegh 2002).
2
Following Carey and Shugart (1998) we distinguish constitutional decree authority
(CDA) from decree authority delegated by congress (DDA). Under spatial assumptions,
delegation of decree powers simply indicates closeness between the president and
congress (in the sense that congress considers the president a trusted agent). We
therefore ignore DDA in most of the discussion below.
3
Formally, the winset W(SQ) can be defined as T U (D
1
U D
2
U...D
J
) U (V
1
∩ V
2
∩...V
K
),
where T is the preferred set of the dictator, D
J
is the preferred set of the j-th decisive
player, and V
K
is the preferred set of the k-th veto player—any of those sets is empty
when the respective type of player does not exist.
4
In fact, the threshold for an override can be thought of as a measure of how much of a
veto player the president is. At the limit, when only a simple majority is required in
congress to override the executive’s veto, the president is deprived of any substantial
veto-playerness.
5
Simultaneously, an increase in the number of individual players presumably expands
the yolk for qualified majorities (q-yolk), which in turn reduces the q-cohesion, and
increases policy stability (Tsebelis 2002, 51-55).
6
These extremes are not merely hypothetical. The first case is illustrated by Colombia,
1958-74, when Liberals and Conservatives constitutionally shared an equal number of
seats in the legislature (ENP=2.00) (Hartlyn 1988); the second, by the Dominican
38
Convention