(Page 1 of 33) - Bargaining in a Spatial Context: An Experimental Study authored by Butler, Christopher. and Bellman, Mary.

Bargaining in a Spatial Context: An Experimental Study

Christopher K. Butler, Ph.D.

Assistant Professor

Department of Political Science

1915 Roma NE, SSCI 2059

Albuquerque, NM 87131

505/277-3742

## email not listed ##

Mary J. Bellman, Ph.D.

Department of Political Science

1915 Roma NE, SSCI 2059

Albuquerque, NM 87131

WORK IN PROGRESS.

PLEASE DO NOT CITE WITHOUT PERMISSION.

Abstract

Spatial modeling is a mainstay of formal theory applied to committees, legislatures, the

electorate, and—more recently—international relations. Bargaining within a spatial context dates
back to Edgeworth in Economics but has often been ambiguous in that numerous points along
the contract curve are equally likely. Part of the difficulty of using an issue space within a
bargaining model is that more specific point solutions are sensitive to various assumptions. Three
well-known cooperative bargaining solutions are the Nash (1950), the Kalai-Smorodonski
(1975), and the Felsenthal-Diskin (1982) bargaining solutions. Each of these only depend on the
structure of the utility space, but each relies on different assumptions, therefore often resulting in
different point predictions, especially within the unidimensional spatial context.

We conducted a face-to-face experiment between thirty-four pairings of men, all

presented with the same bargaining problem in different contexts. For each pairing, we collected
data on a trial bargaining period under a “classic” Nash bargaining problem (under private
information, but without monetary incentive) and two bargaining periods with an underlying
unidimensional issue space (one under private information, the other under full information, each
with monetary incentive). While the Nash bargaining solution was consistently chosen under the
“classic” Nash bargaining problem, it was rarely chosen when the bargaining problem
represented an underlying unidimensional issue space (under private or full information). For
the unidimensional bargaining problem, private information did not facilitate producing any
consistent outcome; under full information, the “equity point”—in which each participant earned
the same dollar amount—was chosen significantly more often than any other outcome.

This research is funded by a University of New Mexico Research Allocations Committee small research grant

(#03-03; IRB Protocol #22166). The authors thank Karin Butler, Randy Calvert, Wendy Hansen, and Kate Krause
for their comments and suggestions on this project. Figures were produced either in MS Excel

or in Stata