External information costs and the adverse selection problem: A comparison of NASDAQ and NYSE stocks

This paper presents an empirical comparison of the bid-ask spread and its components on the NYSE and NASDAQ markets. Using the model of Huang and Stoll (1994) the effective bid-ask spread is decomposed into an adverse selection component and a dealer's gross profit component. We hypothesize that NASDAQ market making firms conduct information search and security analysis which reduces or eliminates the informational disadvantage for NASDAQ dealers. NYSE specialist firms do not typically engage in such analysis, and NYSE regulations prevent the transfer of information to the specialist from the analyst if the firm does perform both functions. The empirical implications of our hypothesis are that the adverse selection component of the spread should be higher on the NYSE than in the NASDAQ market. Additionally, because dealers' gross profit must cover the costs of information search and security analysis, this component of the spread should be higher in the NASDAQ market than on the NYSE. Our empirical results are strongly consistent with the stated hypotheses. On the NYSE the adverse selection component of the spread is significantly higher than that for NASDAQ stocks and it increases monotonically with trade size as suggested by the adverse information literature. For NASDAQ stocks the adverse information component of the spread is near zero and it is unrelated to trade size. The gross profit component of the spread is significantly lower on the NYSE than in the NASDAQ market for trades of all sizes, which is consistent with higher information search and security analysis costs in the NASDAQ market. Hopefully the analysis presented here will contribute to an understanding of how the design of market structures impacts the ability of market participants to deal with the problem of adverse information.