Continuations, Reversals, and Adverse Selection on the Nasdaq and NYSE/AMEX

We show that estimates of adverse selection in the bid-ask spread obtained using methodologies based on Glosten (1987) can be expressed as a weighted average of adverse selection calculated using continuations (successive trades in the same direction) and reversals (successive trades in opposite directions). For example, the methodology of George, Kaul, and Nimalendran (1991) implicitly excludes continuations, using only reversals to assess the degree of adverse selection. Most importantly, though Nasdaq firms exhibit essentially the same degree of adverse selection in reversals and continuations, we find that adverse selection based on reversals far exceeds that based on continuations for NYSE/AMEX firms. These results suggest that reversals and continuations are fundamentally different and the difference is related to market structure. Finally, we argue that continuation based measures of price responsiveness are a natural measure of an important cost of accumulating positions over t..