Does the cost of borrowing shares of a stock for shorting predict its future returns? In their January 2014 paper entitled “The Shorting Premium and Asset Pricing Anomalies”, Itamar Drechsler and Qingyi (Freda) Drechsler investigate shorting fees as a predictor of stock returns. For analysis, they sort stocks at the end of each month into equally weighted tenths (deciles) based on their shorting fee and then examine average future performance of the deciles, both gross and net of shorting costs. They also analyze how shorting fees affect returns to seven known stock return anomalies: value-growth, momentum, idiosyncratic volatility, composite equity issuance, financial distress (likelihood of bankruptcy), max return, and net stock issuance. Using monthly stock shorting fees aggregated across a large number of participants in the stock loan market (from Markit Security Finance), monthly stock returns and firm characteristics for a broad sample of U.S. stocks during January 2004 through October 2012, they find that:
- Over the sample period, for the eight deciles with the lowest shorting fees, the average annual stock borrowing cost is about 0.32%. The average annual borrowing cost in the ninth (tenth) decile is 0.78% (5.8%). The average annual borrowing cost for the 5% of stocks with the highest shorting fees is 9.1%.
- There is a strong non-linear negative relationship between shorting fee and next-month return. Average gross next-month returns are about the same across the eight cheap-to-borrow deciles (0.74% to 0.92%), but drop sharply for the more costly to borrow ninth (0.52%) and tenth (-0.71%) deciles.
- A portfolio that is each month long (short) the decile of stocks with the lowest (highest) shorting fees generates an average gross monthly return of 1.45%, with associated gross four-factor (market, size, book-to-market, momentum) monthly alpha 1.55%.
- Accounting for the cost of borrowing the shorted decile reduces average monthly long-short return to 0.92%.
- Further restricting the short side of the portfolio to the 5% of stocks with the highest shorting fees boosts gross average monthly return (alpha) to 2.13% (2.27%).
- Over the sample period, shorting fees interact strongly with the returns for the seven stock anomalies identified above.
- The anomalies essentially do not exist among the 80% of stocks with the lowest shorting fees, but are very strong among stocks with the highest shorting fees.
- For example, the average gross monthly return for the idiosyncratic volatility anomaly is 0.88%, but this return falls (rises) to -0.04% (1.85%) among the 80% (7%) of stocks with the lowest (highest) shorting fees.
- A five-factor model of stock returns that adds a shorting fee factor to the market, size, book-to-market and momentum factors largely explains all seven anomalies. For example, the average gross monthly four-factor (five-factor) alpha for idiosyncratic volatility is 1.99% (0.38%).
- Analysis of a longer 1980 through 2012 sample based on estimated stock shorting fees (total short interest divided by number of shares held by institutions) generates similar results.
In summary, evidence indicates that stocks with very high shorting fees strongly underperform on a gross basis, and this underperformance accounts for many stock return anomalies. In other words, shorting fees consume much of the gross performance of many reported anomalies.
Cautions regarding findings include:
- The sample period is short, especially relative to anomalies based on infrequent measurement of firm characteristics.
- While the study considers returns net of shorting fees, it does not address an ultimate net that accounts for the trading frictions involved with monthly portfolio reformation. These trading frictions would materially reduce returns for the shorting fee anomaly.
- As noted in the study, high-fee stocks tend to be small, so market capacity to exploit the shorting fee anomaly may be modest.
See also “Lendable Share Supply a Roadblock to Shorting Strategies?”.
