In this paper, we analyze the class of all smooth separating sequential equilibria in a continuous-time bargaining model with two-sided uncertainty. Trade between players occurs whenever there is surplus to be shared and delay is used to signal their valuations. When the buyer and the seller have a common discount rate, we show that the final outcome is unique among all these equilibria: the difference between the highest possible buyer's valuation and the lowest possible seller's valuation always narrows down at a rate exactly equal to the discount rate. When their discount rates differ, the more patient side always reveals his valuation first in the unique smooth separating equilibrium. Key words: Bargaining, signaling, delay, two-sided uncertainty.

I show how uncertainty about agents' future costs of delay, can lead to substantial bargaining delays when agents have reputational concerns. Reputational concerns arise because with positive probability agents are behavioral types, committed to demanding a fixed share of the surplus. In the unique equilibrium, rational agents may demand almost the entire surplus and then wait, with the deadlock only broken by the arrival of news about future costs, even as the probability of behavioral types vanishes. Although each agent would benefit from a compromise reached immediately, they do not propose such agreements because doing so increases an opponent's option value of waiting.

These 22 contributions to the economic theory of non co-operative bargaining show how incomplete information, small numbers of agents, and the rules governing negotiation interact to cause inefficiency, indeterminacy and delay in bargaining outcomes.

We experimentally study unanimity and majority voting rules in multilateral bargaining environments with stochastic future surplus. In these settings, reaching agreement when expected future surplus is sufficiently higher than the current surplus is inefficient. Theoretically, such inefficiencies never arise under unanimity rule but can arise under majority rule as players try to avoid endogenous risk of being excluded from the winning coalitions in the future. We find strong support for this prediction both when the unanimity rule is predicted to lead to more delays, and when both rules should lead to identical levels of delays. We also find that there are more delays than predicted under the majority rule. Using data from conversations among the bargainers and the type of proposals that are implemented, we find that these deviations arise as a result of more egalitarian sharing than predicted by theory, and therefore, lower risk of being excluded from the winning coalitions in the future.