MARKET ORGANISATION AND TRADING RELATIONSHIPS  Ge Ârard Weisbuch, Alan Kirman and Dorothea Herreiner In this paper we give a theoretical model of buyers' behaviour on a market for a perishable good where no prices are posted. We show that if buyers learn from their own previous experience there is a sharp division between those who learn to be loyal to certain sellers and those who continue to `shop around'. This feature remains in more general models which are simulated and is consistent with empirical data from the Marseille ®sh market. Many markets are characterised by trading relationships. Individuals system- atically trade with particular partners in certain markets whilst in others no such stable links are observed. Other markets exhibit a mixture of stable links and `searching' behaviour. Yet the way in which such organisation develops and its economic consequences are not considered in most standard theor- etical models. In a Walrasian equilibrium, for example, the following questions are left unanswered: · How do agents get the information about who demands or supplies which good at what price? Who determines those prices? · How is that information used to determine who will make which transac- tion with whom, thereby clearing the market at each stage and determining market organisation in the long run? One of the objectives of this paper is to examine a situation in which individuals set prices and the way in which those who wish to buy, at those prices, become matched with those who wish to sell. In standard search models (see e.g. Diamond (1989)), 1 buyers sample sellers according to some rule and buy from the cheapest. All sellers are anonymous and are searched with equal probability. There is no memory of where favourable opportunities were found in the past. Such models seem to be plausible for transactions which take place infrequently. Yet many markets are ones on which individuals trade frequently with each other. This is particularly true for markets for perishable goods. Furthermore, in this case, since sellers cannot hold inventories, they only supply the quantities they expect to sell during one session. The essential risk, in our stylised context, for a buyer is not that of paying too higher a price but The Economic Journal, 110 (April ), 411±436. # Royal Economic Society 2000. Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA. [ 411 ]  We thank Derek Smith for help with the Tk/Tcl interactive display of simulation results, Rob Deboer for the use of his GRIND software, Annick Vignes who originally collected the Marseille ®sh data, Paul Pezanis and Marlene Mueller for their help in its analysis, Kenneth Arrow, Larry Blume, Paul David, Steven Durlauf, Fang Fang Tang, Douglass North, Ariel Rubinstein and Nick Vriend for helpful comments, and Olivier Chenevez, Bernard Derrida, Jean Pierre Nadal and Jean Vannimenus for helpful discussions and important suggestions. The Laboratoire de Physique Statistique of ENS is associated with the CNRS (UA 1306). This work was started during a visit by GW and AK to the Santa Fe Institute (with the help of a NATO Collaborative Research Grant CRG 95 1261) which we thank for its hospitality. 1 For more elaborate models, see e.g. Fisher (1973) and Lesourne (1992).