Does the limited supply of lendable shares substantially inhibit successful short selling? In the November 2013 draft of their paper entitled “In Short Supply: Equity Overvaluation and Short Selling”, Messod Beneish, Charles Lee and Craig Nichols examine the profitability of shorting U.S. stocks based on the supply of shares available for lending. They note that the short interest ratio (SIR), the ratio of shares shorted to total shares outstanding, masks the importance of lendable supply. SIR may be low either because few investors have negative views, or because the supply of lendable shares is small. They focus on a proprietary measure of stock lendability from Data Explorer Limited called the Daily Cost of Borrowing Score (DCBS), which ranks stocks from 1 (low cost) to 10 (high cost) based on data collected from a consortium of more than 100 institutional lenders. They define stocks with DCBS of 1 or 2 (3 or greater) as easy/cheap (hard/costly) to borrow. They apply basic findings to assess the realism of short-side returns for the following nine published trading strategies:
- Gross profitability
- Asset growth
- Investment‐to‐assets (underperformance of stocks that overinvest)
- Net operating assets (underperformance of stocks with high net operating assets)
- Total accruals
- Payout percentage
- Net quarterly profitability (underperformance of stocks with low quarterly net income divided by assets)
- Financial distress (underperformance of stocks with high probability of bankruptcy)
- Probability of fraud
Using prices, accounting data and lendable/borrowed shares data for stocks representing about 90% of U.S. equity market capitalization during July 2004 through October 2011 (88 months), they find that:
- On average:
- 13.7% (86.3%) of stocks are hard (easy) to borrow at any given time.
- Stocks that are hard (easy) to borrow have less than 10% (about 20%) of outstanding shares available for borrowing.
- Less than 30% of high-SIR stocks are hard to borrow.
- Stocks that are hard to borrow tend to underperform the market, and stocks that are hardest to borrow exhibit the worst future returns.
- Regarding interaction of lendability and SIR:
- Both extremely high-SIR and extremely low-SIR stocks tend to be hard to borrow.
- SIR has marginal power to predict returns for hard-to-borrow stocks.
- High-SIR stocks that are easy to borrow do not underperform the market.
- Low-SIR stocks that are easy to borrow outperform the market by 0.8% per month.
- Short-side gross returns for the nine trading strategies listed above are significant only among hard-to-borrow stocks. In other words, the short-side returns are effectively unavailable.
- Results based on gross three-factor (market, size, book-to-market) alphas are similar.
In summary, evidence on costs/feasibility of establishing short positions suggests that: (1) SIR is of little use for identifying short positions; and, (2) anomalies that rely heavily on the short sides of portfolios may not be exploitable.
Cautions regarding findings include:
- Reported returns are gross of trading frictions. Incorporating reasonable trading frictions would reduce these returns.
- As noted in the paper, SIR may still be useful in identifying stocks likely to outperform (low-SIR, easy to borrow).
Frequently Asked Questions
What is the Daily Cost of Borrowing Score (DCBS) and how does it show whether a stock is hard to borrow?
The Daily Cost of Borrowing Score (DCBS) is a proprietary Data Explorer Limited measure that ranks stocks from 1 (low cost) to 10 (high cost) to borrow, using data from a consortium of more than 100 institutional lenders. Stocks with a DCBS of 1 or 2 are classed as easy/cheap to borrow, while those with a DCBS of 3 or greater are classed as hard/costly to borrow. It captures the supply of lendable shares, which the short interest ratio does not.
Do short-selling anomaly strategies still work once stock lendability is taken into account?
Short-side returns for nine published trading strategies are significant only among hard-to-borrow stocks, so they are effectively unavailable to investors. The study covered stocks representing about 90% of U.S. equity market capitalization from July 2004 through October 2011 (88 months). The findings suggest anomalies that rely heavily on the short side of portfolios may not be exploitable.
How is the profitability of shorting stocks tested against the supply of lendable shares?
The researchers used prices, accounting data and lendable/borrowed share data for stocks representing about 90% of U.S. equity market capitalization from July 2004 through October 2011. They sorted stocks into easy-to-borrow (DCBS of 1 or 2) and hard-to-borrow (DCBS of 3 or greater) groups, then examined returns by lendability and short interest ratio. They also applied the split to the short sides of nine published strategies, including gross profitability, asset growth, total accruals and financial distress, with gross three-factor alphas giving similar results.
Is the short interest ratio a good substitute for measuring how hard a stock is to borrow?
The short interest ratio (SIR) is not a good substitute for lendability, because a low SIR can reflect either few bearish investors or a small supply of lendable shares. Less than 30% of high-SIR stocks are hard to borrow, and both extremely high-SIR and extremely low-SIR stocks tend to be hard to borrow. High-SIR stocks that are easy to borrow do not underperform the market, so SIR is of little use for identifying short positions.
What are the limitations of the findings on lendable share supply and short-selling returns?
The reported returns are gross of trading frictions, so incorporating reasonable trading costs would reduce them. Only about 13.7% of stocks are hard to borrow on average, yet the short-side anomaly returns are significant only within that group. SIR may still be useful for identifying stocks likely to outperform, since low-SIR, easy-to-borrow stocks outperformed the market by 0.8% per month.