34
significance level of .099. Thus, veto powers seem to be more effective when the
president has a majority of the seats in the legislature. Decree powers were insignificant
in both previous tests, and it is possible that they only have a significant effect when the
president has a high level of partisan support in the majority, so to guarantee the decree
will not be overturned. I test hypothesis 10 by interacting presidential majority with a
dummy variable indicating the presence or absence of a presidential decree power.
Model 7 shows that the interaction variable is significant but that the coefficient
has a negative sign. The variable decree dummy is significant, albeit at a fairly low level
(the p-value is .092), with a positive sign. This means that when the president has a low
level of support in the legislature, decree powers have a positive effect on structural
reform. When the president has a high level of support, decree powers have a negative
effect on reform. Consequently, the results for decree powers are somewhat confusing.
They do not have any consistent strong effect on structural reform. While presidential
partisan powers seem to be a more direct tool for influencing policy, decree powers
certainly should have a more marginal impact. More specifically, it is likely that they are
only used when the president has little partisan support and must rely on other ways to
influence policy. Thus, although theoretically there should be a lower risk of decrees
being overturned with a high level of presidential influence in the legislature, there is less
of a need for decrees in such situations.
In fact, the weak significance of the non-interacted decree variable in Model 7
seems to indicate that decree powers are more likely to be used when the president has a
low level of partisan support. On the other hand, the same logic would predict that decree
powers would not be needed when the president has a high level of partisan support so
Convention