Gains Versus Costs in Legislative Bargaining

Outcomes under the Baron-Ferejohn (1989) model are investigated when proposers distribute benefits versus imposing costs under an experimental design where predicted outcomes are theoretically isomorphic, absent reference dependent preferences. Initial experimental sessions showed greater proposer power under Costs than Gains, contrary to what would be predicted absent reference dependent, or Kahneman and Tversky, type preferences. Questionnaire responses suggested that this greater proposer power under Costs resulted from concerns about being left out of the winning coalition, thereby wiping out their entire endowment. A second set of sessions increased voters' endowments under both treatments resulting in no difference in proposer power over the last five bargaining rounds, consistent with standard expected utility maximizers. This resulted from increased proposer power under Gains, with no change under Costs. The increased proposer power under Gains can be explained in terms of the “numerosity adaptation effect”, widely reported in the psychology literature