top of page

Working Papers:

We study commitment strategies in adversarial bargaining—situations in which the proposer demands payment from the responder under the threat of a conflict. The bargaining environment is risky; a welfare-destroying conflict might randomly start during the negotiation, but it also can be induced by the proposer. If the conflict starts, the responder's loss depends on the scale of the conflict the proposer chooses, and it is maximized for a larger scale than the one that maximizes the proposer's benefit. The proposer has pre-commitment power to a scale given by audience costs. We describe under which conditions the proposer commits to start the conflict to threaten the responder and when she relies on the risky environment as a threat. We also show that having high pre-commitment power can be detrimental for the proposer because of a commitment trap. The proposer prefers a risky environment if she has high pre-commitment and a safer one in the opposite case.

I study how voluntary disclosure of information affects outcomes in plea bargaining. A prosecutor negotiates a sentence with a defendant who privately knows whether he is guilty or innocent. The prosecutor can gather evidence regarding the defendant's type during negotiations, and a trial assigns payoff depending on the evidence if they fail to reach an agreement. Voluntary disclosure induces endogenous second-order belief uncertainty. I show that a purely sentence-motivated prosecutor might disclose exculpatory evidence and that voluntary disclosure generates inefficient outcomes. Mandatory disclosure is socially preferable as outcomes are fairer and efficient. The prosecutor is better off under mandatory disclosure.

We study how commitment affects threats in adversarial bargaining—situations in which a proposer demands payment from a responder under the threat of a conflict if the demand is rejected. The conflict's payoffs depend on the proposer's election of the conflict's scale. To threaten the responder, the proposer chooses the conflict's scale before making the demand. The responder's loss of the conflict is maximized at a large scale than the proposer's benefit. If the conflict starts, the proposer can reduce the scale up to a limit given by the commitment power. Higher commitment implies a lower possible reduction. We test whether, if the commitment power allows it, the proposer chooses the scale that maximizes the responder's loss after the reduction. The experimental data shows that participants with enough commitment power used it to gain bargaining power. Still, the reduced conflict scale is lower than the responder's loss maximizing scale.

I study the role of the negotiation protocol in two-issues bargaining between two players, in which the pie only exists if both players contribute to its creation. The issues are the fraction of the pie, and the second is the pie itself, modeled as which project to choose. I examine three protocols–simultaneous, sequential, and incomplete bargaining–and show that the protocol does not play any role if the contribution cost of one player is high enough. I provide conditions under which the protocol plays a role and describe the equilibrium. I apply the model to discuss the allocation of control rights in early projects financed by venture capital.

I study a bargaining model between two players with endogenous probability of recognition and surplus. At each period they can make two types of effort: productive effort, that increases the surplus, and unproductive effort, which affects the probability of being recognized as the proposer. With convex effort cost players increase the surplus for some periods before ending the game. I characterize how the advantages of each player affect the effort decisions over time. I show that advantages in the unproductive effort affect the provision of both types of effort, but advantages in the productive effort only affect the effort decision regarding the productive effort. Different time preferences only affect productive efforts if the probability of recognition is not persistent, and both types of effort if it is.

bottom of page