On experimental equilibria strategies for selecting sellers and satisfying buyers

We consider marketplaces where buyers and sellers iteratively encounter to trade. Given some specific trade conditions, the question that we address is what strategies should buyers and sellers use to maximize gains. We focus on electronic markets where supply shortages are common. Under such market conditions sellers can only satisfy a subset of the purchase orders they receive from buyers. Consequently, some buyers may become discontented and they may be motivated to migrate to other sellers in the proceeding encounters. Beneficial purchase-order selection as well as seller selection require, respectively, seller and buyer strategies. Analytical computation of stable profiles of such strategies is infeasible in the environments we examine. We hence devise a new methodology for studying strategic equilibria. We introduce specific equilibria strategy profiles to be implemented by automated trade agents. The main conclusions of our study are that automated sellers will benefit most by randomly selecting the purchase orders of their buyers to be satisfied. Additionally, such sellers will not benefit from learning the buyers' typical order size. Moreover, automated buyers will maximize their benefits by re-issuing purchase orders with sellers that satisfied them, fully or partially, in the past.