Assistant professor at UVA
We study a model where a belligerent depends on continued support from an ally to fight a war. Both parties have aligned preferences in war, but the donor prefers to withdraw support if the prospects of winning are too slim. To regulate their ally's beliefs about their ability to win, the recipient may deviate from optimal battlefield strategies in equilibrium. This can occur even when the recipient is strong and victory is likely. In fact, greater fighting ability can make it more difficult for the strong belligerent to fight optimally because setbacks are even more likely to imply weakness. Allowing the donor to occasionally observe conditions on the ground can make both strong and weak recipients worse off. Limited commitments also do not resolve the problem created by the conditionality of support. For any finite commitment length, a sufficiently long war between patient states induces the same behavior as a game with no commitment.