ECONOMY

844 international conflict
and defeat in the war is assured if all revenue is retained for use as private rewards
to coalition members. Suppose V = v + r where v is the public goods component of
victory enjoyed by everyone in the society and r represents the additional resources
(i.e. the spoils of war) that can be extracted from the vanquished state and allocated
to the leader’s coalition members following victory. Defeat has a value of 0. Letthe
per capita cost of waging war be k whether the war is won or lost. If a leader makes
an all-out effort to ensure victory, then the public and private benefits of winning
are worth v + r/W − k for each member of her coalition. Alternatively, if the leader
chooses to retain resources for use as private benefits to her coalition at the expense
of losing the war for sure, then the payoff to her coalition is R/W − k. Theleader
prefers additional effort when v>(R − r )/W, the latter term of which is decreasing
in W provided that R > r ; that is, national resources (R) exceed the value of postwar
spoils (r ). We see, then, that small W systems—like autocracies and juntas—are
more likely to induce leaders to choose to make relatively little extra effort to win a
war, and large coalition systems, like democracies, are relatively more likely to induce
leaders to give up paying their coalition private rewards in exchange for relatively
more extra, marginal effort put into achieving victory in war (Bueno de Mesquita
et al. 1999, 2004).
Empirical evidence reveals that democrats need victory relatively more than autocrats
to survive in office and so make the extra effort described above (Bueno de
Mesquita and Siverson 1995; Bueno de Mesquita et al. 2004). This implies that leaders
of large coalition systems are more selective than autocrats about the circumstances
under which they are prepared to fight, and, further, helps explain why they almost
always win the wars they initiate. In fact, democracies win 93 per cent of the wars they
initiate while autocrats win only about 60 percentofthetime(ReiterandStam2002).
The cases of observed war outcomes, of course, represent a biased sample as many
disputes are resolved through negotiation. The above summary statistics presumably
reflect the situations in which autocrats thought (a) they had an acceptable chance
of victory; the dispute was over (b) a policy concession, such as deposition of the
rival, that the autocrat could not agree to; or over (c) a proposed concession—like
the autocrat going into exile—that the democrat could not credibly commit to honor
ex post. This commitment problem, of course, is the issue faced by Chile’s deposed
dictator Augusto Pinochet; and it may have been an issue in Saddam Hussein’s
decision to fight rather that negotiate a deal with the United States in 2003.
Allowing for the small advantage gained by striking first, autocrats basically have
even odds of winning when they start a war while for democrats victory is practically
certain. But if two democrats are at loggerheads, then war is unlikely. The reason is
that each democratic leader has similar incentives, including an incentive to try hard
if war ensues. Each must provide policy success in order to be retained by his or her
constituents. The likelihood is practically naught that leaders of two rival democracies
each believes at the same time about the same dispute that their prospects of victory
are nearly certain. When democrats do not think they are nearly certain of victory
they opt for negotiations over fighting. Thus leaders of two democracies are unlikely
to find that the circumstances are right for them to gamble on war over negotiations.