The size of this Core is simply a function
of how far apart the House
and Senate median members are. If they are relatively far apart (as with
S
2
and H
3
in
fig. 6
B), the Core is relatively large; if they are close together
(in
fig. 6
C, these two medians are identical), the Core is small.
A Party-Free Presidential System
Finally, consider a presidential system lacking disciplined parties: there is
a bicameral legislature consisting of a House and Senate, plus a presi-
dent. In this system, each of these three institutional actors has author-
ity to block efforts by the others to change policy; the status quo policy
can be upset only when a House majority plus a Senate majority plus the
president can agree on some other policy. Our goal is to determine the
set of equilibrium policies in this system. We assume there is no veto
override.
7
For example, in
figure 7
A the president is at P, and the set of equilib-
rium policies thus spans the ideal points of P and H
3
; that is, the Party-
Free Bicameral Executive Veto Core is
the set of points from P to H
3
.
The reason is that an SQ to the left of P could be upset since there ex-
ists a policy at or to the right of P that the president, all senators, and all
representatives prefer to the SQ. Similarly, an SQ to the right of H
3
could be upset since there exists a policy at or to the left of H
3
that the
president, all three senators, and a majority of House members (H
1
, H
2
,
and H
3
) would prefer to the SQ. But no SQ lying in the P to H
3
region
could be upset; for example, an SQ at H
1
could not be upset by any pro-
posal to its left because all the House members (including H
1
) would
vote against it, and this SQ at H
1
could not be upset by some proposal
to its right because the president and all senators (and H
1
as well) would
vote against it.
If the president is more centrally located, we can get a different Core.
For example, with the president as shown in
figure 7B a somewhat
smaller Core is produced. Note in
figure 7B that the ideal point of the
president could be moved anywhere in between S
2
and H
3
(i.e., within
the bicameral core, ignoring the president) without changing the size or
location of this Core at all.
Depiction of the Party-Free Bicameral Executive Veto Core can be
simpli
fied considerably. First identify the relative locations of the ideal
points of the president, the median House member, and the median Sen-
ate member; for instance,
figure 7A can be reduced without any loss of
Veto Points in Democratic Systems
93
information to what is shown in diagram C, and
figure 7B can be re-
duced without any loss of information to what is shown in
figures 7C
and 7D. The Core is simply the set of policies spanning the median ideal
points of the two “outside” actors. (From this perspective, we are con-
sidering the ideal point of the president to be his institution’s own “me-
dian.”) Thus, in
figure 7C the outside actors are P and H
3
, and so the
Core
spans these points; in
figure 7D, the outside actors are S
2
and H,
and so the Core spans these two points.
The size of this Core depends on the extent of preference differences
among the “outside” pair of actors in the simpli
fied representation (as in
figs. 7C and 7D). If the outermost two actors have similar median ideal
points, the Core will be relatively small. In
figure 7E, for example, the
president, median House member, and median Senate member have
identical ideal points, thereby producing a single-point Core; this depicts
what may be characteristic of an extreme case of “uni
fied party” govern-
ment. But if the outermost pair of actors have rather different median
ideal points, the Core will be relatively large, as in
figures 7F and 7G;
these are two possible types of “divided party” government.
Summary
While we have presented results for just six of the many possible systems,
some important generalizations can be drawn from these systems that I
think are representative of all possible democratic systems. In the next
section, I discuss these generalizations.
Are There System-Related Differences in Core Sizes?
The core for each of our six democratic systems has now been identi
fied.
For any pair of systems, if one system always has a core that is larger than
the core of the other system, we could then conclude that a bureaucracy
in the
first system will always have more autonomy than a bureaucracy in
the second system. We could even arrange our six systems in decreasing
order of bureaucratic autonomy (the smaller the cores the less the auton-
omy), and we could then base an empirical investigation on this expected
(i.e., hypothesized) rank ordering of systems.
However, if for any pair of systems it is not the case that one system
always has a core that is larger than the core of the other system, then we
may be unable to draw any general conclusions about which system’s bu-
reaucracy will have more autonomy. That is, it may be that, for one pair
Veto Points in Democratic Systems
95