configurations and focus on the overall impact of the selected variables. The simulation
results suggest that, according to the logic of the veto-player theory, we should expect
constitutional decree authority and veto override by simple majorities to have a
significant negative impact on policy stability. In both cases, a “decisive player” is born,
but the impact of these institutions can be substantively different because, in contrast to
congress, the executive faces no problem of collective action. In our tests, CDA
increased the size of the winset between 18 and 22 percent, and simple majority override,
by about 11 percent. Against hypotheses 2 and 3, the impact of reactive powers on policy
stability was feeble.
Tables 2 and 3
The number of legislators (a variable of some theoretical relevance but rightly
ignored by empirical studies) showed the expected sign in both specifications but had no
significant impact on the dependent variable. Not surprisingly, the greater the distance
between the mean legislator and the status-quo, the greater the size of the winset. In the
fully specified model (Model 8), the average distance between legislators has a somewhat
intuitive effect. In spite of the larger “yolk” of the collective veto player, greater
separation between legislators tends to create greater policy stability—presumably
because the status-quo is more likely to be in equilibrium.
The simulation challenged the conventional wisdom that a larger effective number
of parties breeds greater policy stability, even after we controlled for instances of
“impossible majorities.” It also suggested that guaranteed majorities can significantly
reduce policy stability as the largest party gains distance from SQ. Against all
expectations, the share of seats of the largest party had no clear influence on the capacity
18
Convention