We consider a network of intermediaries facilitating exchange between buyers and a seller. Intermediary traders face a private trading cost, a network characterizes the set of feasible transactions, and an auction mechanism sets prices. We examine stable and equilibrium networks. Stable networks, which are robust to agents' collusive actions, exist when cost uncertainty is acute and multiple, independent trading relationships are valuable. A free-entry process governs the formation of

more »

... m networks. Such networks feature too few intermediaries relative to the optimal market organization and they exhibit an asymmetric structure amplifying the shocks experienced by key intermediaries. Welfare and empirical implications of stable and equilibrium networks are investigated. We thank Shachar Kariv and numerous seminar and conference audiences for valuable feedback. Leister thanks Sanjeev Goyal and the Economics Faculty at Cambridge University for hospitality. We are grateful for the referees and the editor for valuable suggestions that challenged us to (successfully) generalize several preliminary results. each trader's private trading cost introduces residual uncertainty regarding intermediaries' demand into the environment. When a trader experiences a negative cost shock, the market's operation is shaken. However, if the web of relationships among intermediaries is sufficiently dense, such shocks minimally impact the market as a whole. If the trading network is locally sparse, a shock's impact is exaggerated and market breakdown may ensue. Paralleling questions 1 and 2, we distinguish between stable and equilibrium trading networks. In our analysis, "stability" refers to a network's persistence and is distinct from the network formation process discussed above. Stable networks are immune to the contractive incentives implicit in networked markets and preserve traders' arms-length interactions. In a stable market, existing traders must not be able to profitably merge together in an attempt to exploit complementarities or to curtail competition. Our stability notion captures this intuition and allows us to isolate the distinct, and sometimes subtle, channels through which the incentives to destabilize an existing network operate. Our model suggests that the gain