When Short Sellers Must Buy and No One Will Sell

In January 2021 GameStop's share price rose more than 1,600% in about two weeks, and the widely reported explanation was that short sellers had been forced to buy shares back.

Between the close of trading on January 11, 2021, and the close on January 27, the share price of GameStop rose by more than 1,600%, according to a report that the staff of the US Securities and Exchange Commission (SEC), the country's securities regulator, published in October 2021.^[1] The explanation most widely reported was that short sellers had been forced to buy shares back, and the report noted that "short selling and calls on social media for short squeezes received a great deal of media attention." A short seller borrows shares, sells them, and owes the same number of shares back later, so the bet pays off only if the shares can be bought back for less than they were sold for. A short squeeze, in the usual account, is what happens when the price rises instead and short sellers rush to buy back what they owe, which pushes the price higher still.

That account raises a question the cases in this essay can answer: if short sellers drive squeezes, does a large short interest, meaning many shares sold short and not yet bought back, show that a squeeze is coming? Behind it lies the essay's key question: what drives a short squeeze, and what brought the pressure down in these cases? The essay tests the usual account, because the SEC staff, who studied the trading itself, did not find short covering to be the main driver of GameStop's rise and warned that "the interplay between shorting and price dynamics is more complex than these narratives would suggest." Their evidence comes at the end. Before it come the Northern Pacific Railway in 1901, the central case, which a financial weekly recorded day by day, and Volkswagen in 2008, a briefer comparison told in the companies' own words.

The test starts with what a short seller owes and what can force a purchase.

A short seller owes back borrowed shares, and a demand from the broker, the buyer or the lender can force a purchase when it cannot be met another way.

A short seller's debt is a debt in shares, so a rising price makes it more expensive to repay but does not, on its own, set a date for repaying it. Buying the shares back to close the position is called covering, and until someone with a claim on the position makes a demand, covering remains the short seller's own decision. The sources used in this essay name three parties who can make such a demand: the broker, the buyer and the lender.

The first is the broker, who holds the short seller's collateral, called margin. When the price jumps, the SEC staff wrote, "short sellers would face margin calls requiring them either to post additional collateral or to exit their position," so a margin call forces a purchase only on sellers who cannot or will not add collateral.

The second is the buyer, under the New York Stock Exchange's rules as described in Sereno S. Pratt's 1903 handbook The Work of Wall Street.^[2] The essay assumes that these rules held in May 1901 (the exchange's 1914 constitution gives the same delivery terms; the 1901 text was not found), uses them only for Northern Pacific, and does not claim they applied in 2008 or 2021; for 2021 it relies on the SEC staff's account of margin calls. Shares sold on the exchange had to be delivered by a set hour, and if they were not, the buyer could have the contract "closed out 'under the rule.'" The exchange's chairman would then "publicly buy in the stock at the best price that can be obtained," and the buyer "has a claim against the seller for the difference." The step was the buyer's choice rather than automatic, so a seller who missed the hour depended on what the buyer decided.

The third is the lender, who "has the right to demand return of the stock." A recalled loan that could not be repaid in shares fell under the same rule, so a lender's recall could end in the same way as a buyer's demand.

All three demands act on a debt in shares, and the total of such debts in one company's stock on a given date is its short interest, which this essay gives as a percentage of all the shares the company has issued, its shares outstanding. GameStop's own record shows what that number can tell and what it cannot.

GameStop's short interest was close to 100% of its shares outstanding from 2019 until early 2021, which showed the risk of a squeeze but not its timing.

According to the SEC staff report, GameStop's short interest "hit 50% of shares outstanding first in 2012 and then again in 2015, 2016, and 2018, before rising even further in 2019," and from then until early 2021 it "hovered around 100%." The same report gives the usual short interest at large non-financial companies as below 2.5% of shares outstanding and found that in January 2021 GameStop was the only stock whose short interest exceeded its shares outstanding. That is possible because a share can be lent more than once: a share lent to one short seller is sold to a buyer, the buyer can lend it to a second short seller, and two short sellers then owe the same share.

The price did not stand still while the short interest was that high. In 2020, the staff report records, the share price fell from "just over $6" at the start of the year to "under $3" in April, and ended the year at "a little under $20." Short interest stayed at about that level through those swings, so a high short interest sat alongside large moves in both directions, and the rise that made headlines came in January 2021.

This is why the size of the short interest could not tell anyone when a squeeze would come. A large short interest measures how many borrowed shares are still owed; it does not say when anyone will be forced to buy them back. The forcing comes from one of the demands described above, and even short sellers who are forced to buy drive the price up sharply only if few people are willing to sell to them. At GameStop, therefore, the short interest showed the risk of a squeeze but not its timing, and the missing part of the explanation lies on the other side of the trade, with the sellers.

A case from 1901 shows that other side, because there the shares for sale were taken up by two groups competing for control of a railway.

In October 2008 Porsche disclosed that it held 42.6% of Volkswagen's ordinary shares and cash-settled options on another 31.5%; two days later the shares closed at a record €945, and no primary source shows how much of that rise came from short sellers buying back.

On October 26, 2008, Porsche Automobil Holding SE said that at the end of the previous week it had held 42.6% of Volkswagen's ordinary shares and cash-settled options on a further 31.5%.^[5] Volkswagen also had non-voting preferred shares, which these figures leave out. An option of this kind is a contract that pays its holder the rise of the share price above an agreed price, and because these options were settled in cash, Porsche would receive money, not shares. The banks that wrote these options owed Porsche any rise in the price, and a bank in that position can protect itself by holding the shares; whether these banks did so, and kept the shares rather than selling or lending them, no source states.

Porsche, which said it intended to hold 75% of the ordinary shares in 2009, gave the short sellers as its reason for the disclosure: there were "by far more short positions in the market than expected," and the disclosure would let them "settle their relevant positions without rush and without facing major risks." The other large holder was the state of Lower Saxony. Figure 2 adds its stake at the end of 2008 to Porsche's report, and the remainder it shows rests on that assumption about the banks, which is explained beneath the figure.

Two days after the announcement, on October 28, Volkswagen's ordinary shares closed at €945.00, the highest close in the company's history.^[6] In the days after the disclosure, Porsche announced that it would settle "up to five percent of the hedging positions," that is, part of its option position rather than shares. Volkswagen attributed the fall in its shares that followed to Porsche's announced settlement of part of its hedging positions together with the worsening economy. On the last trading day of 2008 the shares stood at €250.00. In March 2016 the Regional Court of Stuttgart acquitted Porsche's former chief executive, Wendelin Wiedeking, and its former finance chief, Holger Härter, of market manipulation.^[7] The verdict is a reminder that a squeeze is not in itself evidence of wrongdoing.

Volkswagen's record thus rests on the companies' own words, and it cannot show how large the short positions were (see the Methodology) or how much of the rise came from short sellers buying back. GameStop is the case in which a regulator's staff examined who was buying.

SEC staff found that short covering likely contributed to sharp rises in GameStop's price in discrete periods, particularly January 22 to 27, but that the weeks-long rise was sustained by positive sentiment rather than by covering.

The SEC staff report did find a squeeze inside GameStop's rise. Staff observed that "during some discrete periods, GME had sharp price increases concurrently with known major short sellers covering their short positions after incurring significant losses," and that at those times "short sellers covering their positions likely contributed to increases in GME's price"; GME is GameStop's stock-market symbol. Their example was the earlier rise from January 22 to 27, when "the price of GME rose as the short interest decreased." One short seller can also be named. Gabriel Plotkin, whose fund Melvin Capital had bet against GameStop, told a US House committee in February 2021 that the fund "started closing out its position in GameStop at a loss, not because our investment thesis had changed but because something unprecedented was happening."^[8]

Staff also measured the buying as a whole: purchases by accounts that held large short positions on January 15, 2021, were, they found, "a small fraction of overall buy volume," with no percentage given, and "GME share prices continued to be high after the direct effects of covering short positions would have waned." Both findings hold because they describe different scales: across the whole period such buying was a small share, but at particular moments, when large short sellers were covering, it likely added to sharp rises.

What, then, kept the price up? The staff wrote that "the underlying motivation of such buy volume cannot be determined; perhaps it was motivated by the desire to maintain a short squeeze. Whether driven by a desire to squeeze short sellers and thus to profit from the resultant rise in price, or by belief in the fundamentals of GameStop, it was the positive sentiment, not the buying-to-cover, that sustained the weeks-long price appreciation of GameStop stock." In plain terms, the staff could not tell why the buyers bought, and allowed that some may have hoped to squeeze the short sellers; but what kept the price high for weeks was buying by investors who were not short, whatever their reason, rather than short sellers buying back.

The rise also did not wait for the covering. The closing price had already jumped from $19.95 on January 12 to $31.40 on January 13, more than a week before the period the staff linked to covering. On January 13, according to the staff report, the number of accounts trading GME that day rose more than sixfold, from 9,220 the day before to 60,515.

The fall that followed was steep. From an intraday high of $483.00 on January 28, the price fell more than 86% by the close on February 5, and that fall, in the staff's word, coincided with brokers' trading restrictions; the report does not say that the restrictions caused it. GameStop, in short, had a short squeeze inside it, mostly in late January, but SEC staff concluded that a squeeze did not appear to be the main driver.

So the question the essay began with can now be answered, case by case.

A short squeeze happens when short sellers who must buy meet a market with almost no sellers. For May 9, 1901, the Chronicle recorded both conditions in Northern Pacific, and for the short sellers who owed shares to the two banking houses the pressure was relieved by the houses' agreement to settle in cash.

The answer has two parts, because a squeeze needs two conditions at once. The first is short sellers who must buy, and that obligation comes from a demand by the broker, the buyer or the lender. The second is a market with almost no sellers.

How the pressure came down differs from case to case. In Northern Pacific the Chronicle wrote that the tension was "relieved only by the agreement" of the two banking houses, an agreement that applied only to the short sellers who had sold to them; the sources do not say what became of the others, and the records of the time do not say why the houses agreed. Volkswagen's own explanation, Porsche's announced settlement of part of its hedging positions together with the worsening economy, concerns the later fall in its price and says nothing about short sellers. At GameStop the essay makes no claim about an easing: the short covering that SEC staff singled out came in late January, but the price had begun rising before it, and in the staff's words prices "continued to be high after the direct effects of covering short positions would have waned."

That leaves the opening question. A large short interest measures how many borrowed shares are still owed; it does not say when anyone will be forced to buy them back. GameStop's short interest had been about as large as all its shares outstanding since 2019, but the short covering that SEC staff singled out came in late January 2021, and in their view it was not the main driver of the rise. In Northern Pacific what the Chronicle recorded was short sellers who had to deliver shares they could neither borrow nor buy, in a market "swept bare of floating stock." No primary source that we found sizes the short positions in Northern Pacific or Volkswagen, so the essay does not claim that their size is unimportant. What the cases show is narrower: the timing of a squeeze comes from a demand that forces short sellers to buy, and its force comes from how few people will sell to them.