Vultures Circling Overhead: Does Short Selling Tell the Future?

Abstract

This reports evidences a lead-lag relationship between securities which experience high levels of short-selling and those that do not. This is based on evidence that short-selling increases the speed with which information, especially negative information, is absorbed into prices. Previous literature mainly focuses on the presence of short-selling and its effect on prices. This study focuses on the magnitude of short-selling and finds a strong lead-lag relationship between returns of stocks that experience heavy short-selling compared to those that experience slight amounts. The relationship conforms to that of Chordia & Swaminthan’s (2000) speed adjustment hypothesis, in that it facilitates the imputation of common information. The relationship is strongest in small illiquid stocks where short-selling aids in the imputation of common information symmetrically and asymmetrically, and reduces as stocks become larger and more liquid. However in extremely volatile markets this relationship suffers. The relationship is robust to various factors including out of sample tests, accounting for size, and accounting for volume. Of note is the finding that short-selling aids in information imputation over-and-above the efficiency attributed to sophisticated investors. This indicates that market maker and uninformed short-sales add to the lead-lag effect.

Keywords: Auto-correlation, short selling, NYSE, lead lag, Cross-autocorrelation, Pilot program, assymetric relationship, information diffusion

JEL Classification: G12, G14