Abstract

This paper describes a sequenceo f fo ur experiments. The hypo theses tested are all info rmed byo neo r mo reo f the afo rementio ned typeso f mo dels. Each experiment buildso n the findingso f the previo us experiment in the sequence. At the same time, each experiment invo lves a different game, thereby allo wing us to study the pheno meno n fro m a varietyo f perspectives. While so meo f the mo dels we examine get impo rtant featureso f the data co rrect, no ne is co mpletely satisfacto ry. What emerges fro m this sequenceo f experiments is a seto f empirical regularities that sho uld pro vide so me useful guidance to the co nstructio no f mo re accurate mo dels. 2. Bolton and Zwick (1995): Repu tation bu ilding versu s self-centered fairness in an u ltimatu m game At first blush, strategic reputatio n building might no t seem to apply to mucho f theo therregarding behavio r weo bserve in the lab, the reaso n being that lab games are typically played amo ng ano nymo us subjects interacting fo r a single playo f the game. Under such co nditio ns, there is little chance fo r a subject to build a reputatio n witho ther subjects. Buto n reflectio n,o ne might co njecture that lab subjects act to build reputatio ns with the perso n who do es usually o bserve their actio ns: the experimenter. So fo r example, subjects might reject a small shareo f the payo ff to avo id lo o king greedy and unprincipled in the eyeso f the experimenter; by rejecting, the subject establishes a reputatio n with the experimenter that might be pro fitable to the subject in the future

Citations