Estimating bid-ask spread components: Specialist versus multiple market maker systems

This paper tests two competing hypotheses concerning the relationship between adverse selection costs on NASDAQ versus specialist-dominated exchanges. We reject the hypothesis that specialist-dominated exchanges have smaller adverse selection costs than exchanges with multiple market makers. We provide direct evidence on the timing differences between closing transactions and quotes as well as evidence on the extent of nontrading on the AMEX and NYSE but cannot reject the hypothesis that adverse selection costs are a function of average transaction size (which is generally larger on the AMEX and NYSE). We also provide insight into institutional differences across exchanges and the ISSM data base.