Mechanism design theory provides a range of possible outcomes of incomplete information bargaining, based on the welfare weights of the participating agents. However, it provides little guidance on how these weights are determined. This paper provides a framework for selecting bargaining weights based on Nash bargaining over ex-ante surplus. The approach involves a two-stage model for bilateral trade with independent private values. At the ex-ante stage, the agents collectively select a trading mechanism among the domain of all incentive compatible, individually rational mechanisms that requires no subsidy. The bargaining interaction between the agents, once private information is realized, is modeled as a game of incomplete information resulting from the mechanism selected. Alternative solutions such as generalized Nash, Kalai, Kalai-Smorodinsky and Rubinstein are also applicable. This framework can be extended to include multilateral trade, bargaining between partners, sanctions, $k$-double auction bargaining.