What determines how trade in a commodity is divided between privately negotiated transactions via “middle men” (dealer/brokers) in a telephone or “dealer market” versus transactions via “market makers” (specialists) at publicly observable bid/ask prices? To address this question, we extend Spulber’s (1996a) search model with buyers, sellers, and price setting dealers to include a fourth type of agent, market makers . The result is a model where market microstructure — the division of trade between dealers and market makers — is determined endogenously. In Spulber’s model, dealers are the exclusive avenue of exchange, and prices are private in the sense that price quotes can only be obtained through direct contact (e.g. telephone calls) to individual dealers. In contrast a market maker can be conceptualized as operating an exchange that posts publicly observable bid and ask prices. In our model buyers and sellers can either trade with the market maker at the publicly posted bid/ask price or they can search for a better price in the dealer market. We show that the entry of a monopolist market maker can be proﬁtable if it has a lower marginal cost of processing transactions than the least eﬀicient middle man in the equilibrium without market makers. If this is the case the entry of a market maker segments the market; the highest valuation buyers and the lowest cost sellers trade with the market maker and the residual set of intermediate valuation buyers and sellers search for better prices in the dealer market. Dealers act as a “competitive fringe” that undercut the bid/ask spread charged by the monopolist market maker. However less eﬀicient dealers are driven out of business. The remaining dealers are still proﬁtable although the entry of a monopolist market maker signiﬁcantly reduces their proﬁts and bid-ask spreads. Thus, entry by a marker maker results in uniformly higher surpluses for buyers and sellers and higher trading volumes. When there is free entry into market making and market makers’ marginal costs of processing transactions tend to zero, bid-ask spreads converge to zero and a fully eﬀicient Walrasian equilibrium outcome emerges.
Rust, John and Hall, George J., "Middle Men Versus Market Makers: A Theory of Competitive Exchange" (2001). Cowles Foundation Discussion Papers. 1553.