Skip navigation
money-markets.jpg Lazy_Bear/iStock/Getty Images Plus

Do Short Sale Costs Explain Anomaly Returns?

Research shows short sellers are informed market participants, and stocks with the highest lending fees for short sellers still earn abnormally low returns.

Research into short-selling consistently finds the practice plays an important role in market efficiency and the useful allocation of capital. That includes the 2017 study “Smart Equity Lending, Stock Loan Fees, and Private Information,” the 2018 study “The Shorting Premium and Asset Pricing Anomalies,” the 2020 studies “Securities Lending and Trading by Active and Passive Funds” and “The Loan Fee Anomaly: A Short Seller’s Best Ideas,” the 2021 study “Pessimistic Target Prices by Short Sellers,” and the 2022 study “Can Shorts Predict Returns? A Global Perspective.”

The research shows that short sellers are informed investors who are skilled at processing information (though they tend to be too pessimistic)—as evidenced by findings that stocks with high short-seller borrowing costs earn abnormally low returns, even after accounting for those fees earned from loaning shares to the short sellers. Thus, loan fees provide information in the cross-section of equity returns. Interestingly, while retail investors are considered naive traders, the authors of the 2020 study “Smart Retail Traders, Short Sellers, and Stock Returns found that retail short sellers profitably exploit public information when it is negative. The theory is that the high costs and the risk of unlimited losses, and the resulting absence of liquidity-motivated short selling, make short sellers more informed than average traders.

Short Selling Costs and Anomaly Returns

Dmitriy Muravyev, Neil Pearson and Joshua Pollet in their September 2022 study “Anomalies and Their Short-Sale Costs” examine the performance of 162 anomalies to asset pricing models found in the literature and their relationship to the cost of short selling. Their analysis is largely out of sample (minimizing the risk of data snooping) because data on borrowing fees was not available until July 2006, and 83% of the anomalies in their sample were based on sample periods ending before 2006. To distinguish the impact of borrowing fees on penny stocks, they dropped stocks with a price below $1 or market capitalization below $50 million. They then used anomaly signals to sort stocks into decile portfolios.

They began by noting that if anomalies are real (they generate alpha both in sample and out of sample), there should be an important limit to arbitrage that prevents investors from exploiting them. They hypothesized that the stock borrowing fees that short sellers must pay to execute their strategy may be that common limit. This trading cost provides a barrier to arbitrage, preventing sophisticated investors from exploiting the apparent mispricing. Following is a summary of their findings:

  • For the 162 anomalies, the bottom decile portfolio, portfolio one, contained 243 stocks on average, with the number of stocks varying from 24 to 404.
  • The borrowing fee was typically small, most commonly 0.375% per year, but was positively skewed. The borrow fee was 0.25% at the first%ile, 0.38% at the 50th, and reached 3% at the 90th%ile. It was 30% at the 99th%ile.
  • The decile one portfolio tended to contain a high%age (22%) of stocks with high borrowing fees (greater than 1% a year). The mean borrow fee for decile one was 2.7% per year.
  • High-fee stocks tended to have negative abnormal returns.
  • Across the 162 anomalies, the average abnormal return of the decile one portfolio was -0.24% per month and was statistically significant at the 1% confidence level (t-stat = -2.94), while the average abnormal return of the decile 10 minus decile one long-short returns was positive and highly significant, driven by the negative return on portfolio one.
  • Omitting the high-fee stocks, the average abnormal return on the decile one portfolio was almost exactly zero (0.2 basis points per month)—the poor performance of the decile one portfolio was entirely explained by the high-fee stocks.
  • While the average return to long-short anomalies earned a significant 0.15% per month before costs (t-stat = 2.93), this average was -0.02% (essentially zero) once portfolio returns were adjusted for the borrowing fees. Moreover, the anomalies were not profitable before accounting for borrow fees if the 12% of all stocks with high borrow fees were excluded from the analysis.
  • The high-fee stocks accounted for the bulk of the returns to momentum and profitability factor long-short portfolios—an investor who has to pay stock borrow fees to sell the stocks short cannot profitably exploit the positive long-short portfolio returns. Borrowing fees also negatively impacted the return to long-short value factor portfolios and long-short investment factor portfolios.
  • While many anomalies persist out of sample, they cannot be profitably exploited due to the costs of borrowing the stock to short it.

Their findings led the authors to conclude: “Short sale costs explain why these anomalies exist despite arbitrageurs’ best efforts to exploit them.” They added: “Most anomalies cannot profitably be exploited by investors who must pay the borrow fees to short sell stocks.” That explains why the performance persists in the presence of sophisticated investors.

Implications of Short-Selling Information

The research on short selling has led some “passive” (systematic) money management firms (such as Avantis, Bridgeway and Dimensional) to suspend purchases of small stocks that are “on special” (meaning the securities lending fees are very high). Dimensional has done extensive research on securities lending. Using data for 14 developed and emerging markets from 2011 to 2018, it found that stocks with high borrowing fees tend to underperform their peers over the short term. Moreover, stocks that remain expensive to borrow continue to underperform, but persistence of high borrowing fees is not systematically predictable. While the information in borrowing fees is fast decaying, it can still be efficiently incorporated into real-world equity portfolios.

Dimensional also found that while high borrowing fees are related to lower subsequent performance, it is not clear this information can be used to make a profit by selling short stocks with high fees. Borrowing fees are just one cost associated with shorting; short sellers must also post collateral, typically at least 100% of the value of the borrowed securities, and they incur transaction costs. In addition, the asset manager’s research shows there is relatively high turnover in the group of stocks that are on loan with high borrowing fees. For example, fewer than half of high-fee stocks are still high fee one year after being identified as such. Excluding them may lead to high costs if buy and sell decisions are triggered by the stocks frequently crossing the high-fee threshold. After considering the trade-offs between expected return, revenue from lending activities, diversification, turnover and trading costs, Dimensional believes that an efficient approach to incorporate these findings into a real-world investment process is to consistently exclude from additional purchase small-cap stocks with high borrowing fees.

Avantis takes a slightly different approach in designing its fund construction rules. It tries to avoid holding securities that tend to have characteristics associated with high short-borrowing fees.

There is one more important point to cover. The high risks and high costs of shorting allow some inefficiencies to persist, explaining the information provided by short sellers. The recent GameStop episode in which retail investors using social media banded together with sufficient capital to engineer a short squeeze by attacking the short positions of well-capitalized hedge funds demonstrated just how risky shorting can be, and had never been experienced and almost certainly was not expected.

Compounding the risks of shorting, as Xavier Gabaix and Ralph Koijen demonstrated in their 2021 study “In Search of the Origins of Financial Fluctuations: The Inelastic Markets Hypothesis,” is that markets have become less liquid and thus more inelastic. Gabaix and Koijen estimated that today $1 in cash flow results in an increase of $5 in valuation. One explanation for the reduced liquidity is the increased market share of indexing and passive investing in general. Reduced liquidity increases risks of shorting. Adding further to the risks is the now-demonstrated ability of retail investors to “gang up” against shorts. The bottom line is that the limits to arbitrage have increased, allowing for more overpricing of “high sentiment” stocks, making the market less efficient. 

Investor Takeaways

A large body of evidence demonstrates that short sellers are informed investors who play a valuable role in keeping market prices efficient—short selling leads to faster price discovery. Fund families that invest systematically have found ways to incorporate the research findings to improve returns over those of a pure index replication strategy. It seems likely this will become increasingly important, as the markets have become less liquid, increasing the limits to arbitrage and allowing for more overpricing. And finally, the evidence demonstrates that you should not own stocks with high borrowing fees. Forewarned is forearmed.


Larry Swedroe has authored or co-authored 18 books on investing. His latest is Your Essential Guide to Sustainable Investing. 

All opinions expressed are solely his opinions and do not reflect the opinions of Buckingham Strategic Wealth or its affiliates. This information is provided for general information purposes only and should not be construed as financial, tax or legal advice.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.