Short Sales And The Return Predictability Of Peer Stocks

The impacts of short selling have been widely examined by numerous studies.Among them,short selling has been found to affect stock price efficiency,the process of price discovery,liquidity,earnings management,and future returns.Nevertheless,existing studies investigating the effect of short selling have largely focused on its direct effect on the targeted firm and have neglected its spillover effect on other related firms.Short selling activity can influence the price movement of other related firms for several reasons.Due to their limited information processing abilities,investors are not always able to process all publicly available information.This causes slow information diffusion across investors and return predictability across assets.In our setting,short sellers are usually viewed as well-informed traders that have greater information processing ability than other less informed investors have.Therefore,when short sellers trade,their trading behaviors will send information signals to other less informed investors.Meanwhile,the superior information exploited by short sellers is also related to other firms,especially peer firms in the same industry.If short sellers indeed exploit valuable industry-related information in their trading,this industry-wide information can be related to other firms in the same industry.This paper studies the impact of short selling activity from a different perspective.Different from previous studies that focus on the impact of short selling on the shorted firms,we are interested in the spillover effect of short-selling activity from short sellers to other investors holding non-shortable stocks.Specifically,we examine whether the industry-level short selling activity measured by the average of short interest across shortable firms in the industry has predictive power for the future abnormal returns of non-shortable firms in the same industry.Relying on portfolio sorting and the Fama-MacBeth regression,we find that the average of short interest in the industry is negatively related to the future abnormal returns of non-shortable firms in the same industry.In the portfolio sorts,the “High-Low” strategy generates statistically significant abnormal returns ranging from-0.65% to-0.88% per month for the value-weighted portfolios.The results from the Fama-MacBeth regression are consistent with that found in the portfolio sorts.The results are robust to a variety of alternative explanations,including the difference in firm characteristics,lead-lag effects,overvaluation caused by short sales constraints,and short sellers' concentration in the industry.Moreover,the predictive ability of industry short selling is not short-lived and can last up to 6 months.To further explore the information content embedded in the industry level short interest,we repeat the analysis in the Fama-MacBeth regression but use the short flow from a randomly selected industry.Not surprisingly,the negative and significant relation between industry short flow and the future returns of nonshortable firms disappears.In addition,we find evidence that the predictive ability of industry short selling is intensified as the expansion of the shortable list in an industry,suggesting more industry information is embedded.Focusing on the newly added firms,our industry short selling survives after the control of firm-level short selling.Last,we find that short selling activity in the industry is negatively related to the future changes of firm fundamentals.