A Better Measure of Institutional Informed Trading

Abstract

AbstractAlthough many studies show that the presence of institutional investors facilitates the incorporation of accounting information into financial markets, the evidence of informed trading by institutions is rather limited in the extant literature. We address these inconsistent findings by proposing PC_NII, percentage changes in the number of a stock's institutional investors, as a novel informed trading measure. PC_NII is better able to detect informed trading than are changes in institutional ownership (ΔIO)—the measure commonly used in previous studies—because (i) entries and exits are usually triggered by substantive private information and (ii) only a small fraction of institutions have superior information. As conjectured, PC_NII subsumes the information content of ΔIO and other institutional trading and herding measures in the forecast of stock returns, and its strong predictive power for stock returns reflects mainly its close correlation with future earnings surprises. We also show that PC_NII helps address empirical issues that require a reliable measure of institutional informed trading.