based on a crisis bargaining game with two-sided incomplete information (section 3),
and test the explanations using the equilibrium solution of the game (section 4).
3
Fi-
nally, I look at individual cases to elaborate the causal mechanism suggested by these
empirical results (section 5).
The empirical results suggest several conditions under which sanctions are more
e¤ective (or ine¤ective). Most of all, the dyadic relationship between the sender and
target turns out to be an important factor that determines the outcomes of sanctions.
The sender gets signi…cant disutility in deadlock situations where the allied target
resists but the sender does not lift the sanctions. Sanctions against the allied targets
are in general incredible in the sense that the sender will prefer back-down to the
continuation of the sanctions if the allied target resists. Allied targets, however, tend to
comply in many cases even though they can anticipate that they can win the sanctions
dispute. Moreover, non-allied targets are likely to resist even though they can expect
that sanctions will not be lifted if they resist. Given that sanctions are more frequently
imposed on non-allied targets, the overall success rates of sanctions should be low.
These results are then elaborated in four case studies. In each case, I focus on the
e¤ect of alliance considerations on the sanctions’credibility and the target’s decisions.
The severity of the issue at stake matters in determining the onset of sanctions.
The sender receives additional utility if the target accepts the stronger demands. This
makes the sender engage in sanctions disproportionately if the objective of the sanctions
is important. For example, sanctions are more likely if the sender aims at destabilizing
the target’s leader than if the demand is a minor policy change. However, issue salience
does not always induce a higher rate of compliance. An increasing number of sanctions
imposed for salient issues increases the concession probability if the sanction’s level is
high, but not if it is low. The target’s domestic institutions also a¤ect the e¤ectiveness
of sanctions. Under a weak level of sanctions, democratic targets tend to be worse o¤
in non-compliance outcomes, thereby increasing the probability of compliance. Finally,
economic dependence of the target fails to achieve statistical signi…cance.
3
I focus on the derivation of outcome probabilities and the estimation of coe¢ cients of interest not
only when we assume states behave strategically but also when states have uncertainty with respect
to one another (Signorino 2003, Lewis and Schultz 2003, Wand 2005).
3
Convention