A comparative test of three models for solving “the bargaining problem”

Three experiments are reported, each of which attempts a comparative test of three models for solving the bargaining problem: those of Nash (1950), Kalai & Smorodinsky (1975), and Felsenthal and Diskin (1982). Human subjects in two‐person bargaining situations were used in each of the experiments. A simple game matrix—named “Send‐the‐Marines” after a scenario which illustrates its basic structure—was used under several conditions, including those of a known series of trials in which parties could agree in advance concerning their responses and conditions of a single play for which side payments could be negotiated. These two sets of conditions produced quite different patterns of results, with the side‐payments‐possible condition showing most clearly patterned agreements. A special factor here was the “exchange bonus,” which came into existence with side payments to extend the effective range of possible agreements. Although some of the results are consistent with one or another of the three models tested, none of the models is consistent with all of the results. A particular problem is that the assumed base for bargaining (the “conflict point” or “minimum utility point” of the models) was used by subjects as a base for producing agreements only under limited conditions. A more general model which includes considerations of the prominence of a possible solution, its social efficiency, and evaluations of equity appears better to describe the full results than any of the formal models examined.