We present a finitely repeated bargaining game with complete information. The stage game is asimultaneous demand game with a fall-back position for both parties, in which we allow one party(say, the union) to estabilish a credible commitment to strike if it is not offered a determinedwage. We try to refine the equilibrium set of the repeated game using a formulation of ForwardInduction. In particular, we say that a path of Nash Equilibria in the repeated game is Consistentwith Forward Induction (CFI) if for all period t the cost of deviation (if it is strictly positive) is greateror equal than the maximal net gain in CFI paths with t-1 horizon. We present several cases in which the average payoff for the union in any CFI path, when thehorizon tends to infinity, is his preferred wage. These results are similar to those obtained with thereputation effects approach and reveal some connection between the FI notion and the approachconsisting of perturbing the game with some incomplete information.
