Does the power of short interest to predict future returns derive from superior information of short sellers or from overvaluation driven by short-selling constraints? In their January 2008 paper entitled “Why Do Short Interest Levels Predict Stock Returns?”, Ferhat Akbas, Ekkehart Boehmer, Bilal Erturk and Sorin Sorescu examine evidence that discriminates between the competing information and overvaluation explanations. Their key discriminators are: (1) the effects of levels of and changes in institutional ownership (availability of shares for shorting) on the predictive power of short interest; and, (2) the relationship between short interest and subsequent news. Using daily stock returns, monthly short interest, quarterly institutional holdings, firm fundamentals and news/earnings reports spanning 1988-2005, they conclude that:
- Over the sample period, heavily shorted stocks underperform lightly shorted stocks by 1.51% (1.72%) monthly on an absolute (risk-adjusted) basis.
- High levels of short interest predict future negative news and negative earnings surprises, and this predictive power is stronger for firms with low institutional ownership.
- A hedge portfolio that is long (short) lightly (heavily) shorted stocks earns on average 0.31% per day during the four days surrounding earning announcements.
- Among stocks with low institutional ownership, a hedge portfolio that is long (short) lightly (heavily) shorted stocks earns an average return of 0.70% per day around earning announcements. This average return declines to 0.04% for stocks with high institutional ownership.
- Stocks with increasing (decreasing) institutional ownership are associated with lower (higher) short interest informativeness.
- Hedge portfolios that are long (short) the most heavily shorted stocks with the largest increases (decreases) in institutional ownership return an average 7.71% per month during the quarter for which the change in institutional ownership is calculated.
In summary, short sellers have acted as specialized monitors who tend to know what they are doing, but high levels of and large positive changes in institutional ownership can obscure short interest informativeness.
This result indicates that relaxation of short-selling constraints may not eliminate the informativeness of short interest.
