High Cost of Shorting and Overvalued Stocks
Short-sellers play a valuable role in keeping market prices efficient, which is important to the efficient allocation of capital. Larry Swedroe unpacks new research into high shorting costs and overvalued stocks.
The U.S. equity lending market, essential for short-selling, has more than $1 trillion in market capitalization on loan, and it accounts for more than a third of the trading volume on domestic exchanges. The securities lending and shorting markets are, thus, important venues for the process known as price discovery.
Typically, stock lenders are institutional investors, such as pension funds, insurance companies and mutual funds, holding large portfolios of securities. They lend to collect fees and enhance income on their stock holdings through fully collateralized loans.
Borrowers are usually hedge funds, arbitrageurs and proprietary trading desks, which short the stock as part of a trading strategy. Lenders generally use custodian banks, with which they split lending revenue, retaining 70-80% of lending fees. Stock borrowers use prime brokers or broker-dealers to locate securities available for lending.
How Shorting Impacts Returns
There are two main views on how shorting impacts the cross section of stock returns. First, from a “market frictions” perspective, short-sale constraints prevent pessimistic investors from trading and, without their views impounded in prices, stocks can become overvalued and subsequently underperform.
Historically, equities with high short-sale constraints have earned lower subsequent returns. This helps explain the existence, and profitability, of stock market anomalies (mispricings).
Second, from an “asymmetric information” perspective, high short interest can represent informed investors with negative private information who identify overvalued stocks and, thus, subsequent underperformers. Again, empirically, this view is also well-supported by evidence that high short interest is associated with low subsequent returns.
It has also been well-documented that the costs of shorting (the borrowing fee paid to securities lenders) depend on utilization, quantity available to lend, shares outstanding, liquidity, turnover, past returns, volatility, size, book-to-market ratio and a number of accounting variables related to asset pricing anomalies. The accounting variables related to anomalies used to estimate short-sale costs are asset growth, gross profitability, accruals, Ohlson O-Score, net operating assets and net investment growth.
Research On Shorting
Brian Henderson, Gergana Jostova and Alexander Philipov recently contributed to the literature on short selling with their May 2017 study, “Shorting Fees, Private Information, and Equity Mispricing.”
They hypothesized that the candidate explanations for shorting costs can be classified into three categories: those related to supply and demand; those related to information uncertainty and asymmetry; and those linked to anomalies and potential mispricing.
The authors examined a number of regression specifications and produced robust estimates of predicted and abnormal fees. Their data sample covers 95% of the market capitalization of CRSP-listed U.S. equities and includes a total of 5,628 firms, with a monthly average of 3,589 firms, from July 2006 to March 2013.
Following is a summary of their findings:
The highest-fee quintile of expensive-to-short stocks has lower prices, smaller market capitalization, higher book-to-market ratios, higher volatility, higher analyst dispersion and forecast bias (i.e., the percentage of forecasted EPS relative to actual EPS), higher Amihud illiquidity, higher credit risk and default probability.
High-fee stocks also exhibit lower institutional ownership, lower breadth of ownership, lower quantities available for lending and higher utilization, consistent with being harder to borrow.
Stocks in the most-expensive-to-borrow quintile report unexpectedly low earnings (negative earnings surprises), while the rest of stocks have positive earnings surprises. However, they are indistinguishable from other stocks in terms of accruals, short-interest ratio or market betas. They have lower (negative) HML (or, high minus low) exposure and more negative momentum exposure than the remaining stocks, and earn four-factor alphas of -80 basis points per month.
Overpricing and subsequent underperformance occur only in the third of expensive-to-short stocks that have high premiums in their lending fees.
Only short-sellers who pay these substantial premiums are profitable, which reveals they trade on private information. Such informed trading represents less than 3% of all short interest (and less than 7% of all stocks borrowed).
Lending fee discounts, alternatively, imply informed lending. Lenders maximize revenue by offering large fee discounts on stocks with the most elastic demand and extracting the highest premiums on stocks with the most inelastic, private-information-driven demand.
Henderson, Jostova and Philipov found lending fees vary considerably across stocks.
They write: “Fees for the most expensive-to-short quintile average 411 basis points (bps) per year, while they range from just 5 bps to 26 bps per year for the four less expensive quintiles. Consistent with the ‘market frictions’ literature, the highest fees (a direct measure of short-sale constraints) are associated with the lowest returns—the highest shorting fees quintile earns -60 bps per month and underperforms the lowest fees quintile by 140 bps per month.”
The authors decomposed the lending fee into a “fair fee” and an “abnormal fee,” using monthly cross-sectional regressions of fees on proxies proposed by the literature for short-sale constraints and market frictions.
They explain: “The fair fee equals the fitted regression value, while the error term represents abnormal fee. Hence, the fair fee incorporates available public information and serves as a proxy for market frictions while lending fee premiums and discounts potentially represent private information. A positive error term indicates that the stock is being lent at a premium above what is justified by fee determinants.”
Henderson, Jostova and Philipov found:
The premiums and discounts on lending fees are substantial.
Among the quintile of stocks with the highest actual fees, one-third have an average premium of 130% over the fair fee and one-third have an average discount of 60% relative to the fair fee.
Stocks in the quintile of highest actual fees are those typically referred to as “special” or “hard to borrow.”
Their new insight was that, while “special” stocks on average earn lower future returns, two-thirds of these hard-to-borrow stocks do not significantly underperform stocks in other fee quintiles, all of which are easy-to-borrow, or “general collateral,” stocks.
This result challenges the ability of “market frictions” to independently explain the negative future returns of high-fee stocks. Only the one-third of the stocks in the hard-to-borrow quintile with high abnormal fees significantly underperformed stocks from any other portfolio.
Specifically, the authors found:
Within the hard-to-borrow quintile, stocks with high abnormal fees underperform stocks with low abnormal fees by 12% per year, and other hard-to-borrow stocks lent at a “fair” fee by 19% per year.
The high-fee-premium, hard-to-borrow portfolio generates annualized returns of -14%, a four-factor risk-adjusted alpha of -23%, and underperforms any one of the other 24 portfolios in the five-by-five sort by 10-32% per year on a risk-adjusted basis.
Henderson, Jostova and Philipov concluded: “This evidence suggests that the high shorting fee-low return relation identified by the market frictions literature can be attributed exclusively to this high premium hard-to-borrow portfolio rather than all stocks with high shorting fees, and that this relation is primarily driven by private information. The stocks in the high premium hard-to-borrow portfolio are likely shorted by informed investors because in the presence of significant short selling restrictions only informed traders will short. Short sellers of these stocks are willing to pay the highest fees in our sample (900 bps per year on average), and they are the only traders who realize net profits from their short positions. As further evidence, the high premiums of these hard-to-borrow stocks are likely associated with private information because we do not find any apparent link between the premiums and company characteristics or any other public information.”
They added: “High premiums in the fees of this small subset of special stocks (6.5% of our sample) represent short sellers’ knowledge that is truly orthogonal to public information.”
Security Lenders Well-Informed
Henderson, Jostova and Philipov furthermore concluded that “stock lenders are also informed, but their private information appears more related to lending-market conditions than to future stock returns. Lenders are smart about offering fee discounts or maintaining high fee premiums in pursuit of revenue maximization.”
The authors found that “the stocks with the highest fee premiums are associated with the most inelastic demand, while their utilization indicates that additional shares are available for lending. This suggests that lenders recognize that additional loans at lower premiums would decrease revenues due to the inelastic demand. On the other hand, difficult-to-short stocks with the largest fee discounts have the most elastic demand as well as the highest utilization, indicating lenders’ willingness to lend as many shares as possible at significant discounts to ‘fair fees’—a strategy that we estimate maximizes lending revenue.”
They continue: “The lenders’ discounts attract significant numbers of uninformed short sellers—the largest fee discounts are associated with the highest supply of lendable shares, the highest utilization (77% of available shares), and the highest short interest ratios. This shorting demand appears uninformed because, even discounted, fees are not low enough for the short side to make a profit.”
Henderson, Jostova and Philipov documented that high short interest can arise in two situations. Both occur among stocks lent at a high premium or at a large discount.
They write: “However, only one of these channels, the one linked to hard-to-borrow stocks lent at high fee premiums, is associated with significant negative future returns. There, short interest appears driven by negative private information about future returns. Alternatively, high short interest may be driven by lending fee discounts, possibly combined with negative public information about the firm’s prospects.”
The stocks most expensive to borrow (at 331 bps) are the past-six-month losers that continue to be losers in the next six months.
By comparison, past losers that become future winners are almost 2.5 times less costly to borrow (150 bps). These are less expensive to borrow than the past winners that become future losers (at 191 bps per year).
For every past return quintile, future losers are at least twice as expensive to borrow as future winners.
The negative returns of abnormally expensive hard-to-borrow stocks persist for three months after portfolio formation, with a cumulative three-month return of -3.51%, and continue to be the only portfolio that generates net shorting profits. It also generates a significant four-factor alpha of -1.90% over the subsequent month and -5.51% over the subsequent three months.
The authors concluded that their results “imply that the majority of short interest is actually uninformed and significantly influenced by discounted stock loan fees as well as by publicly available information related to trading frictions.”
They add: “This evidence is not fully consistent with a market frictions explanation that high fees, a proxy for short-sale constraints, limit the shorting of pessimistic investors and lead to inflated prices. Rather, the pattern appears more consistent with an asymmetric information explanation that investors with negative private information about the firm’s future prospects may bid up the borrowing costs of potentially overvalued stocks.”
Finally, they state: “Premiums in lending fees of hard-to-borrow stocks are likely driven by negative private information about the company’s future prospects. Short sellers are willing to pay substantial premiums for a reason their trades are the only profitable short positions in our sample. Despite paying the highest fees (82 bps per month), these short sellers earn gross shorting profits of 117 bps per month. They are the only short sellers who generate a net positive shorting profit, supporting the hypothesis that high fee premiums proxy for private information. As further support for this hypothesis, we do not find that these stocks differ in their firm characteristics, or any other public information, from other special stocks.”
Short-sellers play a valuable role in keeping market prices efficient, which, in turn, is important to the efficient allocation of capital. If short-sellers were inhibited from expressing their views on valuations, securities prices could become overvalued and excess capital would then be allocated to those firms.
Henderson, Jostova and Philipov provide important new insights on shorting fees, showing both that abnormal fees are a stronger predictor of negative future returns than short interest, and that large negative returns arise primarily among some special stocks, and are accompanied by high fee premiums bid up by a small subset of informed short-sellers.
The authors also show that lenders are smart about maximizing their revenue from lending special stocks. They discount lending fees and extend supply for stocks with the most elastic demand, while maintaining high premiums and withholding supply for the stocks with the most inelastic demand.
Their analysis of shorting supply and demand showed that fee discounts appear to be associated primarily with lenders, while fee premiums appear to be associated with short-sellers. These results are consistent with the interpretation that the majority of short interest is uninformed and primarily influenced by negative public information related to trading frictions and already embedded in stock prices.
Finally, they show that high short interest and utilization, typical proxies for short-sale constraints, may not necessarily be linked to higher lending fees or subsequent negative returns.
The research on short selling has led some “passive” money management firms (such as AQR, Bridgeway and Dimensional Fund Advisors in its long-only funds) to suspend purchases of stocks that are “on special” (that is, where securities lending fees are high). (In the interest of full disclosure, my firm, Buckingham strategic Wealth, recommends AQR, Bridgeway and DFA funds in constructing client portfolios.)
This commentary originally appeared September 15 on ETF.com
By clicking on any of the links above, you acknowledge that they are solely for your convenience, and do not necessarily imply any affiliations, sponsorships, endorsements or representations whatsoever by us regarding third-party Web sites. We are not responsible for the content, availability or privacy policies of these sites, and shall not be responsible or liable for any information, opinions, advice, products or services available on or through them.
The opinions expressed by featured authors are their own and may not accurately reflect those of the BAM ALLIANCE. This article is for general information only and is not intended to serve as specific financial, accounting or tax advice.
© 2017, The BAM ALLIANCE