before replacement, and the probability of conflict may be greater than in previous periods (recall
that either challenger or defender may take escalatory steps in this model). This also represents a
secondary novel prediction; lame duck leaders in democracies may under some conditions be partic-
ularly likely to become involved in conflicts. However, since the functional form of the relationship
is nonlinear in the defender’s probability of survival (from Proposition 6), Chiozza and Goemans’
(2003, 2004b) nonstrategic research design may be inappropriate for uncovering the true relation-
ship. Further, the empirical analysis of strategic interaction benefits greatly from formalization,
and this paper presents a first step in that direction.
• New leaders bargain “harder” than leaders with more time in office.
This result resolves the question of which leaders bargain “harder” than others (Chiozza and
Goemans 2003, 2004b; Gelpi and Grieco 2001), also without reference to questions of tenure-
inefficiency. New leaders simply stand more to gain from overstating their resolve than experienced
ones for two reasons. First, if they develop a reputation for resolve, they expect ceteris paribus to
remain in power longer and to receive a longer stream of benefits. Second, for those leaders that
engage in bargaining, the reduction over time of opponents’ uncertainty around their true resolve
limits the size of potential changes in an opponent’s beliefs.
14
Put simply, developing a useful
reputation early is better on balance than trying to make up the same ground later.
• New leaders who successfully bluff achieve better bargaining outcomes over time than those
who unsuccessfully bluff or who honestly reveal low resolve.
While both Chiozza and Choi (2003) and Huth and Allee (2002) find apparent reputational
effects on the choices of strategy in and outcomes of territorial disputes, they diverge on specific
findings. Part of the problem, as discussed above, is the choice of disputes as the unit of analysis,
14
This challenges the argument that leaders avoid “diversionary targets,” although it does not
rule out the possibility that some tenure benefits do exist off the equilibrium path as a result.
28
Convention