respectively. All the other countries (Argentina, Bolivia, Chile, Colombia, Panama,
Uruguay and Venezuela) exhibit an average duration ranging from 658 days (1.8 years) to
1,211 days (3.3 years).
Based on the figures listed above, one can propose some suggestive associations. In
the first place, countries with less fragmented legislatures and frequent single-party cabinets
tend to have longer lasting governments, and the opposite holds for countries with high
legislative fragmentation and frequent coalition governments. These findings are fully
consistent with those in the literature on parliamentary regimes (e.g., Lijphart 1999) and
presidential ones (e.g., Deheza 1997).
Mention also should be made that, with the exception of Costa Rica, the United
States, Mexico and Venezuela, there is great intra-national variation in the duration of
presidential cabinets. The countries in which such variation is observed are nearly all, save
Argentina, ruled by coalition governments. It is thus tempting to say that the countries
characterized by the so-called coalition presidentialism do not feature a single pattern of
governance. These countries tend to have diverse governing formulas, also marked by
variable performances. Nevertheless, only a multivariate econometric test can tell us, with a
greater margin of safety, what the most significant determinants of cabinet durability in the
Americas are. This is what I proceed to do below.
Regression analysis
To estimate models of cabinet durability, I use the standard Cox technique (Box-
Stenffensmeir and Jones 2004; Cox 1972; 1975). The advantage afforded by this
econometric technique is that one does not need to assume any probability distribution of
cabinet survival.
19
Convention