ETFs, broadly defined as exchange-traded portfolios of assets, have proven to be among the most successful financial innovations of recent history. Since the introduction of the first ETF in the early nineties, the ETF market ballooned globally to an estimated size of over $6 trillion. Despite this success, regulators and practitioners often express concerns over the potential adverse effects of ETFs. These concerns are not unfounded—ETFs can increase volatility and harm the liquidity of underlying securities. However, the evidence is inconclusive, and many gaps in our knowledge remain.

In a recent paper, entitled ‘Short Selling Equity Exchange Traded Funds and its Effect on Stock Market Liquidity,’ we examine ETFs from a new angle. Using the setting of the US 2008 short-sale ban as a laboratory, we ask whether ETF shorts could have been used to substitute for stock shorts, circumventing the ban. Our results show that the answer is yes; traders used the largest, most liquid ETF, the S&P 500 Spider (ticker symbol SPY), to successfully bypass the ban. Existing empirical literature finds that short-selling constraints, including the 2008 ban, are severely detrimental to stock market quality. Hence, we also investigate whether ETF shorts can alleviate the adverse effects of short-sale constraints. We find that the ETF shorting loophole partially shielded the ETF’s constituent stocks from the adverse liquidity effects of the ban.

In an attempt to stabilize the turbulent financial markets, on September 18, 2008, the US Securities and Exchange Commission uncharacteristically imposed temporary short-selling restrictions on 797 financial-sector stocks, which were subsequently lifted on October 8, 2008. However, no ETFs were listed as restricted. This unexpected short-sale ban provides researchers with something close to a natural experiment, which can be used to study important links between asset classes from a short-selling perspective.

The short interest of the Spider increased, on average, by 35% during the ban period. We estimate that, at its maximum, around $5 billion of new short positions were established using this ETF during the ban. In addition, we document that, during the ban, the number of Spider shares outstanding increased by around 26%, which can be seen as evidence of the ‘create-to-lend’ practice (creating ETF shares for the sole purpose of lending them to short-sellers). Our findings are in stark contrast to those of the prior literature, which shows a contraction in all other bearish trading strategies.

The patterns discussed above suggest that the increase in SPY short sales was driven primarily by investors circumventing the ban. Although the Spider was not a perfect substitute, once one considers its characteristics vis-à-vis other available ETFs and the relevant institutional details, the SPY emerges as the most appropriate instrument to bypass the ban. First, the SPY is well-established, large, liquid, and resilient. It is the US’ oldest and the world’s largest, most liquid ETF. Second, short-selling the Spider as a way of bypassing the ban allowed short-sellers to mask the real intent of their trades and to minimize the risk of their new short positions also being banned. These were short-sellers’ key concerns at the time, due to regulatory uncertainty and regulators’ intimidation tactics. Third, shorting the Spider provided effective short exposure to the banned financial-sector stocks, as around 70 stocks in its portfolio were banned stocks. We also entertain an alternative hypothesis that this surge in SPY short sales was driven by a sudden increase in market pessimism. However, we find no evidence to support this explanation.

Financial theory predicts, almost unanimously, that regulatory short-sale constraints deteriorate liquidity. Given that Spider constituents could be sold short indirectly via the ETF short sales, they were relatively less short-sale constrained than other, similar banned stocks. In line with theoretical predictions, we find that, while the average liquidity of the banned stocks severely deteriorated during the ban, the detrimental liquidity effect was around 30% less severe for Spider’s banned constituent stocks.

Policy implications

Our work shows that ETFs can be used to indirectly short-sell stocks. This mechanism can alleviate short-sale constraints and improve liquidity. It may also be particularly relevant in settings where short sales are banned. Our findings have direct implications for policymakers and market practitioners because short-sale bans remain widely used policy tools, despite extensive evidence of their adverse effect on market quality. For example, in response to the extreme market volatility at the onset of the COVID-19 crisis, short-selling bans were introduced in Austria, Belgium, France, Italy, Indonesia, Greece, Malaysia, South Korea, Spain, and Turkey. Given the global proliferation of ETFs, any future regulation would require a more elaborate design to restrict short selling of the underlying assets whilst leaving the workings of the ETF market undisrupted and less easily exploited.

Egle Karmaziene is an Assistant Professor of Finance at Vrjie Universiteit Amsterdam.

Valeri Sokolovski is an Assistant Professor of Finance at HEC Montréal.