Abstract
In real markets, price adjustment and the matching of buyers and
sellers involve considerable exchange of information. Previous
experience is also important in partner selection and in the decision
to accept a transaction. The dynamic processes involved can be
described in terms taken from cognitive science such as learning and
recognition. We have attempted to model such organization in markets
for perishable goods. Our model is based on preferential choice of
sellers by buyers according to fidelity parameters which are specific
to buyer and seller pairs. Fidelity parameters depend upon the previous
history of transactions between partners. We have shown that depending
upon how much they care about previous profits in selecting their
trading partners, buyers become either faithful partners or searchers.
The transition between the two behaviors resembles phase transitions in
statistical physics. Analysis of data from the Marseille wholesale fish
market confirms this clear distinction between the two possible
behaviors and its relation to past profits.