Research by Jean-Edouard Colliard and Thierry Foucault Published in the Journal of Finance
Many important financial assets (e.g., bonds, derivatives, currencies) trade in decentralized markets, often referred to as “over‐the‐counter” (OTC) markets. In these markets, dealers play a key role because they intermediate trades between end‐investors. To do so, they accumulate substantial inventory positions. These positions are costly to hold and are well‐known to have a significant bearing on market liquidity. To minimize their inventory costs, dealers trade among each other in the inter‐dealer market. However, dealers in OTC markets are typically heterogeneous. While some are well‐ connected, others have only few trading connections, and thus may find it more difficult to adjust their portfolio in the desired direction.
The paper by Jean-Edouard Colliard, Thierry Foucault and Peter Hoffmann proposes a model of trading that studies the joint effects of dealers’ connectedness and inventory costs on prices and allocations. Consistent with stylized facts, we assume a two‐tiered structure with “core” and “peripheral” dealers. While trade is frictionless in the core, peripheral dealers bargain over the price with a limited set of other peripheral dealers. Importantly, only some of them are “connected” and can trade with core dealers.
Given the structure of the model, there are two sources of market power among peripheral dealers. First, connected dealers have the option to resort to trading in the core, which improves their bargaining position relative to unconnected dealers. Second, a dealer’s inventory position relative to that of his/her competitors matters. In a market that is mainly populated by buyers, a seller will find it easy to get his/her trade done. In contrast, a buyer will have a hard time finding a seller and may thus not be able to trade. Importantly, we show that there is price dispersion in the periphery (a sign of some traders exerting market power over others) if, and only if, both frictions are present, i.e., if some dealers are unconnected and, at the same time, if there is an aggregate imbalance between buyers and sellers. “Our model makes precise predictions concerning how empirical researchers can measure the value of connectedness in OTC markets, a topic that has deserved considerable attention in the literature”, explain the researchers.