Other factors that Kellam’s research suggests are relevant to explain cabinet
stability are federalism (federal countries have less stable cabinets) and the degree of
ideological polarization in the legislature (the higher the polarization, the less durable
cabinets are).
In this paper, the approach will be similar to that of Deheza (1997) and Kellam, i.e.,
the dependent variable is the duration of the party makeup of cabinets. This decision results
from the criteria for defining a new presidential cabinet used in Amorim Neto (2006). A
new presidential cabinet is defined by the taking office of a new president or a change in
the party makeup of the cabinet, that is, the entry or exit of any party necessarily implies
the formation of a new cabinet.
Some authors use other criteria to identify a new presidential cabinet (Almeida
2003; Deheza 1997). For example, the holding of midterm congressional elections (because
they alter the legislative makeup and consequently the backdrop for bargaining for cabinet
spots), or a change in 50% or more in the identity of cabinet officers (because a change of
that magnitude signals that the president wants or has to substantially modify political
relations and/or government policies).
These two criteria fail for two reasons. First, having to consider all the events that
can alter the bargaining conditions and/or external conditions of cabinets produces an
endless list of such events. So, as pointed out by Laver (2003, p. 26), a new criterion must
be included in the definition of a new cabinet – the merger of two parties, since such a
decision, just like an election, will affect the legislature’s party composition. In the case of
Brazil, where the rate of party switching by lawmakers is extremely high, a phenomenon
2
I have already used the latter criterion in a work on cabinets in Brazil (Amorim Neto 2002).
9
Convention