Dynamic Adverse Selection and Liquidity

Abstract

Does a larger fraction of informed trading generate more illiquidity, as measured by the bid-ask spread? We answer this question in the negative in the context of a dynamic dealer market where the fundamental value follows a random walk, provided we consider the long run (stationary) equilibrium. More informed traders tend to generate more adverse selection and hence larger spreads, but at the same time cause faster learning by the market makers and hence smaller spreads. These two effects offset each other in the long run. In a more general setup with public news, the offsetting result depends on the persistence of news.

Keywords: Learning, adverse selection, dynamic model, stationary distribution

JEL Classification: G14, D82