Masterarbeit, 2020
59 Seiten, Note: 1,0
List of Figures
List of Tables
List of Abbreviations
List of Symbols
1 Introduction
2 Short Selling (Bans)
2.1 The Idea of Short Selling
2.2 Evolution of Short Selling (Bans)
2.3 European Short Sale Bans in 2020
3 Literature Overview
4 Stock Liquidity
4.1 Variables and Data
4.2 Methodology
4.3 Empirical Results
4.4 Robustness Check
4.5 Effects of the Ban Lift
4.6 Country-by-Country Analysis
5 Price Efficiency
6 Stock Prices
7 Conclusion
Appendix
Bibliography
I investigate the effects of short sale bans on the market microstructure of European equity markets. In Spring 2020, Austria, Belgium, France, Greece, Italy, and Spain imposed short sale bans for a period of two months as a response to increasingly volatile financial markets driven by the Covid-19 crisis. I use daily stock data from 28 European countries to assess the impact of these bans on market liquidity, price efficiency, and stock prices. The analysis of stock liquidity measured by the percentage bid-ask spread, the Amihud illiquidity ratio, and the dollar volume points to a severe degradation in stock liquidity of large firms compared to the control group. I apply a matching procedure to verify the robustness of these results. Reverse effects take place once the ban is lifted again. Furthermore, my empirical evidence suggests that short sale bans reduce the price efficiency as proxied by a price delay measure. However, this effect is not stronger during market downturns. Moreover, it seems that the bans do not affect the speed of stock prices adjusting to firm-specific information and, contrary to the predictions made by Miller (1977), short sale bans appear to have no effect on stock prices. Overall, the results of my study using data from the Covid-19 crisis in 2020 mostly confirm the findings of previous studies that have examined the impact of short sale bans on the market microstructure.
Ich untersuche die Effekte von Leerverkaufsverboten auf die Marktmikrostruktur europäischer Aktienmärkte. Als Reaktion auf die durch die Coronakrise bedingten hochvolatilen Finanzmärkte führten Österreich, Belgien, Frankreich, Griechenland, Italien und Spanien im Frühjahr 2020 Leerverkaufsverbote für einen Zeitraum von zwei Monaten ein. Ich nutze tägliche Aktiendaten aus 28 europäischen Ländern um den Einfluss von besagten Leerverkaufsverboten auf die Marktliquidität, die Preiseffizienz und die Aktienkurse zu untersuchen. Die Analyse der Liquidität mittels der prozentualen Geld-Brief-Spanne, der Amihud Illiquiditätskennzahl und des Dollar-Handelsvolumens deutet auf einen Rückgang der Liquidität großer Firmen im Vergleich zu der Kontrollgruppe hin. Ich nutze ein Matching-Algorithmus, um die Robustheit der Ergebnisse zu verifizieren. Nachdem das Verbot aufgehoben wird, treten entgegengesetzte Effekte auf. Des Weiteren deuten meine empirischen Ergebnisse darauf hin, dass Leerverkaufsrestriktionen die Preiseffizienz, gemessen anhand einer Preis-Verzögerungsvariable, reduzieren, wobei dieser Effekt in fallenden Märkten nicht stärker ist. Zudem scheinen die Leerverkaufsverbote keinen signifikanten Einfluss auf die Anpassung der Aktienpreise an firmenspezifische Informationen und, anders als von Miller (1977) prognostiziert, keinen Einfluss auf die Aktienkurse zu haben. Zusammenfassend kann ich mit meiner Studie anhand der Daten aus der Coronakrise im Frühjahr 2020 die Ergebnisse bisheriger Studien zur Wirkung von Leerverkaufsverboten auf die Marktmikrostruktur weitestgehend bestätigen.
Figure 1: Overall Median Liquidity
Figure 2: Liquidity Proxies over Time
Figure 3: Mean Liquidity Differential over Time
Figure 4: Liquidity Proxies over the whole Period
Figure 5: (Cumulative) Abnormal Returns
Appendix Figure 1: Mean Liquidity Proxies over Time
Appendix Figure 2: Median Liquidity Differential over Time
Appendix Figure 3: Histogram
Appendix Figure 4: Market Return
Table 1: Summary Statistics
Table 2: Median Liquidity
Table 3: Liquidity Analysis Regression Output
Table 4: Liquidity Analysis Quartile Regression Output
Table 5: Robustness Check Regression Output
Table 6: Robustness Check Quartile Regression Output
Table 7: Effects of Ban Lifts
Table 8: Country-by-Country Estimates
Table 9: Price Efficiency
Table 10: Bull-and Bear Markets
Table 11: Abnormal Returns
Appendix Table 1: Liquidity Analysis Quartile Regression Output
Appendix Table 2: Robustness Check Quartile Regression Output
Appendix Table 3: Price Efficiency
Appendix Table 4: Bull- and Bear Markets
Appendix Table 5: Computation of Sub-Samples
Appendix Table 6: Weekly Stock Returns
Appendix Table 7: Abnormal Returns
ADR American Depository Receipt
AMEX American Stock Exchange
CDS Credit Default Swap
ETF Exchange-Traded Fund
EU European Union
NASDAQ National Association of Securities Dealers Automated Quotations
NYSE New York Stock Exchange
OLS Ordinary Least Squares
SE Standard Errors
SEC Securities and Exchange Commission
TARP Troubled Asset Relief Program
UK United Kingdom
US United States of America
Abbildung in dieser Leseprobe nicht enthalten
On March 12, 2020, the EURO STOXX 50, the major European stock index declined by more than 12% - the largest loss ever reported on a single day since index inception in 1986.1 Other major stock indices around the globe experienced a similar drop. The Covid-19 crisis was about to hit and there was great uncertainty among investors, reflected by very volatile stock markets. The corresponding volatility index VSTOXX had its peak on March 16, 2020, with an implied volatility of 86%.
To stabilize capital markets and restore the confidence of investors during volatile times, regulators can make use of temporary bans on short sales, i.e., restricting investors in their ability to profit from declining stock prices. This was the case in numerous countries during the financial crisis in 2007-09. National authorities prohibited short selling either only on financial stocks or all stocks. For example, the Securities and Exchange Commission (SEC), the national regulator for capital markets in the United States, banned short sales on 797 financial stocks in fall 2008. Another example is the short sale ban imposed by four European countries on August 11, 2011, as a reaction to the European debt crisis and the subsequent volatile capital markets.
As a response to extremely volatile markets in March 2020 caused by the Covid-19 pandemic, national authorities for financial markets in Austria, Belgium, France, Greece, Italy, and Spain prohibited covered short sales on all stocks in their country, respectively. The Italian regulator announced the ban to be active for three months while the other five national authorities initially set a ban period of approximately one month which was later extended until May 18, 2020. Following the other countries, Italy also announced to lift the ban on the same day. Thus, the short sale ban was in effect for two months in all six countries and ended on May 18, 2020.
In this paper, I empirically investigate the impact of short sale bans on the market microstructure of equity markets by focusing on three major stock market characteristics: (How) did the ban affect stock liquidity, (how) did the ban influence price efficiency, and did the regulators manage to stabilize stock prices by imposing short sale bans? The first two questions shall be analyzed to examine whether the bans had any undesired side effects such as a decrease in market quality or market efficiency. The latter one is of interest to the extent that price stabilization is one of the primary goals of regulators when restricting short sales. Herewith, I contribute to the existing literature by using the most recent data from the Covid-19 crisis in spring 2020 which caused six European countries to impose bans on covered short sales. I use both daily and weekly data from 28 countries on firm-level for a time span of six months to conduct my analysis.
My empirical results suggest that overall market liquidity decreased drastically in March 2020, but it did so even stronger for large stocks that were subject to the short sale ban. The effect rebounded once the short sale ban was lifted again. These findings are mostly consistent with previous literature and robust against the identification strategy of the diff-in-diff structure as well as the type of control sample.2 Furthermore, I find evidence suggesting that prices of banned stocks adjust more slowly to market-wide information than a set of non-banned control stocks, which is also in line with previous findings. My results, however, suggest that this effect is not stronger during market downturns. Moreover, stock prices appear to remain unaffected by short sale bans, suggesting that regulators could not artificially push stock prices upwards.
As of the start of my thesis, I was not aware of any research dealing with the same topic that had already been published. However, in the meantime, two working papers by Siciliano and Ventoruzzo (2020) and Della Corte et al. (2020) have been published on SSRN which also investigate the effects of short sale bans, but with a less comprehensive approach regarding the data and number of variables to proxy liquidity.
The remainder of this paper is structured as follows: In section 2, I introduce the idea of short selling, its historical evolution, and the corresponding regulatory measures against it in six European countries in Spring 2020. Section 3 presents a review of the three literature streams on stock liquidity, price efficiency, and stock prices. Section 4 elaborates on the liquidity analysis containing a description of the data, the constructed variables, and corresponding descriptive statistics. Furthermore, I explain the methodology, my empirical approach, and provide detailed results including a test for robustness. In section 5, I analyze the price efficiency during short sale bans and present my results. Section 6 provides the results of a stock return analysis in the context of short sale bans, followed by the overall conclusion in section 7.
In this section, I describe the mechanics and general idea of short selling as well as the history of short sale bans before I provide a brief overview of the European short sale bans introduced in Spring 2020.
If an investor enters a long position in any kind of financial instrument, she will gain if the price of the financial instrument increases. The most straightforward type of a long position is to buy stock in a publicly traded company. The opposite happens if an investor enters a short position in any kind of financial instrument. In that case, she will gain if the price of the financial instrument decreases. If an investor possesses the stock and wants to profit from potentially declining prices, she can simply sell the stock and buy it back after the stock price has fallen. The difference between the price at which the stock was sold and the price at which the stock is bought back equals the gross profit of the investor. However, if an investor does not possess the stock, but wants to profit from potentially falling prices, she can enter a short position.
As I use public equity data for this paper, I focus on short positions in stocks. There are multiple ways to enter a short position in a stock. An investor can borrow the stock from a lender (often a market dealer) at a certain fee. Usually, the borrower of the stock must pledge some collateral. The investor then sells the borrowed stock at the stock market, e.g., at a price of €100. After the stock price has fallen, she buys it back, e.g., at €90, and returns the stock to the lender. In this case, the gross profit for the short seller equals €10.3 This is usually known as classic short selling.
Another way of entering a short position is to use derivates, i.e., financial instruments whose value depends on some underlying value (the stock price in this case). An investor can profit from a declining stock price by “selling” a futures contract on that stock. Thereby, she commits to deliver the stock on a certain date in the future receiving the forward price, e.g., €100. If the stock price has considerably fallen between the date of the contract’s sale and the settlement date, and trades now at, e.g., €90, the investor will gain €10 because she buys the stock at €90 at the open market but receives €100 from the counterparty.4
Another potential derivative strategy is a put option. This gives the investor the right, though not the obligation, to sell a stock at a pre-determined price, the so-called strike price or exercise price, e.g., €100, to the option writer. If the stock trades at €90 shortly before the put option expires, the investor can buy the stock at the open market for €90 but sell it to the option writer for €100 at maturity, thereby earning a gross profit of €10. As the name already indicates, the buyer of a put option can but does not have to exercise the option whereas both counterparts of a futures contract are obliged to deliver when the contract expires.5 Hence, the investor must pay a premium upfront for an option whereas a futures contract is simply a deal between the two parties.6 The short sale bans which I investigate in this paper prohibited investors from entering new- or increasing existing short positions by any of the means described.
There are several risks associated with short selling. The most straightforward one is the risk of unlimited losses. If an investor enters a long position by buying a stock, the largest possible loss she faces equals the stock price as prices can (theoretically) not become negative. However, when entering a short position, the potential loss is (theoretically) unlimited as the stock price could increase infinitely. Furthermore, share lenders have usually the right to call in shares at any time. When this is the case, the borrower must return the borrowed shares. To do so, the borrower must either find another lender to roll-over the loan or close the position by buying the shares at an inconvenient price at the open market.7
There are five main reasons for market participants to enter short positions. First, investors such as hedge funds simply speculate on declining prices to make profits. They might have negative (private) information about a single firm that is not yet incorporated in the stock price but will be in the future. This is referred to as informed short selling. Boehmer et al. (2008) argue that institutional short sellers are usually very well informed, as heavy short positions taken by these informed institutions result in a strong risk-adjusted outperformance over the following month. This outperformance is strongest for very small stocks, which is not surprising given the public availability of data on small firms, to which institutional investors have certainly better access. Second, investors use short positions to hedge their portfolios and to steer the overall portfolio risk in accordance with the investment guidelines. Third, market makers enter short positions to provide liquidity to the overall market. If a market maker receives a buy order from an investor but does not have enough stock in the inventory, she simply borrows shares and sells them to the investor. High-frequency traders provide a substantial share of market liquidity and are therefore also referred to as informal market makers, which are not officially registered.8 Market making accounts for the largest part of short selling activity. Officially registered market makers are generally exempt from short sale bans because otherwise, liquidity would deteriorate drastically.9 Fourth, some investors use short selling to exploit market inefficiencies. For example, they sell the stock and buy a convertible bond of the same firm. This is referred to as arbitrage.10 A fifth reason to enter a short position is a quantitative investment strategy. Over the years, many stock market anomalies have been discovered and published. They represent a large literature stream in asset pricing. One famous example is the study by Jegadeesh and Titman (2001) who document an outperformance of stocks that have exhibited high returns in the previous period (week or month), also referred to as momentum. Another anomaly was published by Sloan (1996) who shows that firms with high accruals underperform firms with low accruals. Quantitative investment strategies based on these anomalies work as follows: The investor buys the stocks in the (usually) highest decile of the anomaly variable distribution, i.e., in these cases the return from the previous period or the accruals, and shorts the stocks in the lowest decile of the distribution. This is referred to as a long-short portfolio, the intuition being that this approach requires an initial wealth of zero. If too many stocks in the short leg of the strategy are subject to a short sale ban, either the strategy no longer works properly, or the short leg includes only stocks that are exempt from the ban. Andrikopoulos et al. (2013) find that, on average, only about 30% of the firms in the short leg can actually be sold short, either due to non-availability or due to a short sale ban. Interestingly, however, they do not find any significant performance difference between a constrained- and unconstrained short leg.
Moreover, investors cannot only enter short positions on single stocks but also on a whole range of firms by simply shorting Exchange-Traded Funds (ETFs).11 I provide a detailed explanation in section 2.3 whether and how ETFs and other index-based instruments were subject to the short sale bans in 2020.
The classic short selling, i.e., selling a share that the seller does not possess, can be split into naked- and covered short selling. Covered short selling implies that a short seller has actually borrowed the shares whereas naked short selling means that the share has not even been borrowed. The latter practice has been uniquely prohibited in the European Union (EU) since 2012 and is therefore not considered in my empirical analysis. The significance of short selling activity is stressed by the data. Wang et al. (2020) report a mean short volume ratio of 36.36% of stocks listed on the New York Stock Exchange (NYSE), the American Stock Exchange (AMEX), and the National Association of Securities Dealers Automated Quotations (NASDAQ) between 2010 and 2015. In other words, more than one-third of the trading volume in this period were short positions (excluding derivatives trading).
In times of market turbulences, the practice of short selling is often blamed for amplifying volatility and adding downwards pressure on security prices. However, many researchers argue that it enhances market efficiency and provides liquidity. I give a review of research findings regarding the effectiveness and side effects of short sale bans in section 3.
Short selling and resulting short selling bans have been common practice for centuries. One of the first reported short sales was conducted in the Netherlands in 1609 by a syndicate of traders led by a former director of the Royal East India Company. They sold forward contracts with a maturity of up to two years and tried to bring down the stock price of the Royal East India Company by spreading bad rumors about the business when maturity approached. Company directors reacted by submitting a petition to the States of Holland, basically requesting a short sale ban on company stocks. The States followed the request by setting restrictions against trading in forward contracts. In the end, the syndicate did not succeed anyway, as they did not manage to push down the stock price. In fact, the stock price rebounded (most likely due to a dividend distribution) after it had fallen, resulting in a significant loss for the short sellers.12 Another rather historical example is the NYSE during World War I. It issued restrictions against short selling as they did not want war effort to be lowered by a demoralized stock market.13 The short sale ban in 2008/09 during the financial crisis was the most extensive ban in this century due to the large number of countries worldwide that implemented temporary measures against short selling either on only financial- or all stocks. On September 19, 2008, Christopher Co, the SEC chairman back then, stated:
“The emergency order temporarily banning short selling of financial stocks will restore equilibrium to markets. ” 14
Like many other national regulators, the SEC found it necessary to undertake this step to stabilize financial markets. Brunnermeier and Oehmke (2014) develop a model according to which short selling can indeed be harmful to financial institutions and justifying why it should be considered to protect these firms from short selling. They argue that speculative predatory short selling and the subsequently declining share price can lead uninsured depositors to withdraw their funds due to the fear of impending bankruptcy.15 In the presence of leverage constraints, financial institutions are then forced to sell long-term assets at fire-sale discounts resulting in a loss. This theoretical model shows that a short sale ban may be reasonable in particular cases. However, Christopher Cox had a very critical opinion about the bans ex-post, as he stated in an interview with Reuters on December 31, 2008:
“Knowing what we know, I believe on balance the commission would not do it again. The costs appear to outweigh the benefits.” 16
This underlines the uncertainty about bans and their impacts on markets that regulators face. Yet, there are still authorities that impose bans in times of volatile financial markets. The ban adopted by four European countries as a reaction to the sovereign debt crisis in 2011 is a famous example from the near past. Bris et al. (2007) state that, although short sellers have been made responsible for market crashes for centuries and despite the ongoing debate between regulators and institutional investors, the effects of short sale restrictions are still ambiguous. Market crashes are indeed associated with short-selling activity as measured by the short interest, i.e., the number of shorted shares divided by the number of outstanding shares. Ofek and Richardson (2003) find that, during the dotcom bubble, the short interest of internet firms was significantly larger than the short interest of control firms, also referred to as old economy firms. However, they do not state that short sellers were the drivers of falling internet stock prices. Callen and Fang (2015) conduct an analysis with US equity data and state that short interest today is positively correlated with crash risk in one year. They argue that short sellers are capable of finding bad news that is not yet published by the management and trade on them.
Overall, this paper seeks to provide an overview of the specific literature streams and to extend previous findings using the latest data from the stock market crash in March 2020 as a consequence of the Covid-19 crisis.
The short sale ban in Spain started on March 17, while it was imposed one day later in Belgium, France, Greece, and Italy and two days later in Austria on March 19. All countries except Italy initially set a period of one month for the ban to be in effect with an additional remark to lift it earlier or to extend it if found to be necessary. The latter was the case as all of these countries eventually extended the ban until May 18. The Italian Authority chose a ban period of three months straight away.
For each of these six countries, the ban applied to shares admitted to the respective national trading venue and for which the national regulator was the relevant competent authority. Two examples illustrate and clarify this setting: Apple is a US-American stock that is listed on an American stock exchange but also on the stock exchange in Milan, Italy. However, since the Italian regulator is not the relevant competent authority for Apple, its stock was not subject to the ban, even though it is traded in Italy. Furthermore, OMV is a stock that is traded on the Vienna Stock Exchange, for which the Austrian regulator is the relevant competent authority. Hence, OMV was subject to the ban. The stock is also traded on Xetra, a German trading venue, where the ban also applied to OMV. The ban implied that investors were no longer allowed to enter or increase net short positions. It also applied to transactions in related securities such as derivatives. Market making activities, specific hedging activities, and transactions in index-related instruments such as ETFs with a weight of banned stocks lower than a certain threshold were excluded from the ban.17 Furthermore, the ban explicitly also applied to short positions that are closed within the same trading day. Despite the different announced ban periods at the start of the ban, the bans lasted until May 18, resulting in a ban period of two months in each of the six countries.
On the other hand, other large European countries did not ban short sales in Spring 2020. This emphasizes that the opinions on short sale bans vary widely. The Financial Conduct Authority, the British regulator, issued an opinion in which they state that there is no evidence for short sales to be the driver of falling markets. They argued that short selling is a crucial provider of liquidity. German authorities had a similar view and hence, did not issue any restrictions on short selling. The German fund association stated on March 20 that they decline a general ban on short selling. The fact that some European national regulators imposed short sale bans while others did not allows me to divide my data into a treatment group consisting of all firms from the six countries with a ban and a control group comprising all firms of the remaining European countries without a ban. I provide a detailed description of the data in section 4.
To understand what kind of research has been done on short sale bans, this section provides an overview of the literature on the effects of short sale bans on stock liquidity, price efficiency, and stock prices. Since the inception of short sale bans is a rare event that usually only occurs during severe financial market downturns, a large share of the empirical papers focusses on the financial crisis in 2007-09. While some papers examine only a subset of the effects, others investigate the whole range of effects. In accordance with the empirical part of my paper, I divide the literature section into three parts: The effects of short-sale bans on stock liquidity, price efficiency, and on stock prices. During the financial crisis in 2007-09, short-sale bans varied from country to country with respect to the ban period, to whether only naked or also covered short sales were prohibited, and to whether the bans applied to all or only to financial stocks. As naked short sales have been banned in the EU since 2012 anyway and there were six countries in the EU that all imposed bans on covered sales on all stocks for an identical period of time in spring 2020, my investigation is somewhat less extensive since there is no variation among these characteristics in my data.
Liquidity Effects. Beber and Pagano (2013) use a global dataset comprising 30 countries from the 2007-09 crisis to examine the effects of short sale bans on stock liquidity. Their results suggest that short-sale bans adversely affect daily bid-ask spreads, which they use as a proxy for stock liquidity. These results remain significant no matter if estimated as a panel regression or one regression per country at a time. The decline in liquidity is strongest for small-cap stocks and stocks without listed options. The latter indicates that investors still use derivatives markets to enter short positions. Despite this they find that, once bans on covered short sales are lifted, liquidity tends to increase again. They verify the robustness of their findings by using the illiquidity measure introduced by Amihud (2002) as an additional proxy for liquidity and by running a Two-Stage Least Squares Regression to account for endogeneity. Boehmer et al. (2013) use a dataset from the United States (US) to investigate the effects of the US short-sale ban in 2008 on trading activity and bid-ask spreads. They report a significant decline in daily trading dollar volume for the largest 25% of the banned stocks. A similar picture arises for different measures of market quality, such as relative quoted and effective spreads or intraday price ranges, which tend to increase for the largest 50% of banned stocks after the ban inception. The results are consistent with Lobanova et al. (2010) who also investigate the effects of the 2008 short-sale ban in the US. They find a sharp increase in spreads and a decline in turnover ratio, trading volume, and dollar volume. Marsh and Payne (2012) investigate the effects of short sale bans by using a dataset from the United Kingdom (UK) from 2008/09. They report a reduction in trading volume and market liquidity once a short-sale ban is in effect. Alternative proxies of market liquidity such as the order book liquidity confirm the findings. Helmes et al. (2017) use an Australian dataset to investigate the short sale ban from 2008/09 in Australia. Their control group comprises matched comparable Canadian stocks with respect to industry and firm size. In line with the other studies, they find a reduction of trading activity and an increase of the bid-ask spread for stocks subject to the ban.
Contrary to all these papers, Appel and Fohlin (2010) find that short sale bans in 2008 had a positive or at least a neutral effect on stock liquidity. Similar to my paper, they consider only bans on covered short sales. Their sample comprises 35 stocks from eight (mostly) European countries. Their identification strategy differs from conventional ones in that they match each banned stock with its corresponding American Depository Receipt (ADR) that is traded on a US-American stock exchange and is therefore not subject to the ban.18 They argue that non-banned ADRs differ from banned stocks only in the short sale regulation and are therefore ideally suited as a control group. Only some non-American firms have ADRs trading in the US which explains the low number of included firms in their sample. This reflects the trade-off between a large sample and an accurate control group.
Price Efficiency. Short selling is often associated with more efficient markets.19 Boehmer et al. (2008) conclude that “short sellers possess important information, and that their trades are important contributors to more efficient stock prices”20. Diamond and Verrecchia (1987) show in their model that, once a short sale ban is active, stock prices adjust more slowly to private information. This effect is stronger for negative information, the intuition being that only those who own the stock can trade on negative news whereas others cannot. The result is that it takes longer for stock prices to incorporate information, especially negative news. This is in line with the intentions of regulators hoping that bad news is not immediately incorporated into stock prices as the impact of bad news might be overrated. As it is hard to capture private information, many papers also focus on market-wide information. This theory has been tested in numerous studies of which most support the theory. For example, Beber and Pagano (2013) entirely confirm the theoretical predictions. They try to capture private information by analyzing the autocorrelation of residuals. They report a higher autocorrelation during a ban, suggesting that it takes longer for firm-specific information to be impounded into stock prices during a ban. To distinguish between negative and positive market-wide information, they compute the downside (upside) cross-autocorrelation, i.e., the autocorrelation between individual stock returns and negative (positive) market returns. They find that this measure significantly increases after the ban enactment, that downside cross-autocorrelations are larger than upside cross-autocorrelations and that this difference is larger for the ban period compared to non-ban periods. Intuitively spoken, this result indicates that it takes longer for negative market-wide information to be impounded into stock prices than it does for positive information, especially during ban periods. Bris et al. (2007) report a significantly higher market model’s downside R² for banned stocks compared to non-banned stocks. Their findings also suggest that the price efficiency of banned stocks decreases due to short sale bans. Saffi and Sigurdsson (2011) use the lending supply of stocks and loan fees to proxy short sale restrictions.21 Their findings are in line with the previously described papers: The higher the restrictions are, the lower the price efficiency is. Reed (2007) finds that when short selling is rather costly, the speed with which prices adopt after quarterly earnings announcements tends to be slow. This effect is stronger for bad news announcements. Chang et al. (2007) use data from the Hong Kong stock market and report that stock prices adjust slower when they are subject to short sale bans.
Stock Prices. Miller (1977) developed a theoretical model for short sale restrictions: His key assumption is that investors have heterogeneous beliefs about the development of security prices.22 If a certain security is banned from short-selling, informed market participants who do not own the security but have a pessimistic opinion about it are not able to incorporate their valuation into the security’s price. This leads to inefficiently overpriced securities above the fundamental value as prices reflect only the opinions of bullish investors and bearish market participants who possess the security. This, in turn, means that prices should decline once a short sale ban is lifted. Contrary to that, Diamond and Verrecchia (1987) argue that rational investors take into account the fact that some bearish investors cannot trade on negative information and adjust their valuation subsequently. This does not lead to upwards biased security prices. Beber and Pagano (2013) investigate whether short sale bans help stabilize stock prices. They argue that this is one of the regulator’s main objectives when imposing short sale bans. They don’t find any evidence for significant positive abnormal stock returns, except for the US. However, they suggest that other publicly available information such as the Troubled Asset Relief Program (TARP) announcements may have had positive effects on stock prices in the US. The authors conclude that the major advantage of the political measures that were taken to restrict short-selling activities is the large amount of data that was generated. Similar conclusions are drawn by Boehmer et al. (2013) who do not find any outperformance of banned stocks compared to a matched sample after controlling for the TARP announcement effects. In fact, they find a negative performance of firms that were added to the ban lister afterward, arguing that a lagged addition to the list is a signal of bad firm performance. Saffi and Sigurdsson (2011) do not find any effects on stock prices, either. Their findings suggest that relaxing short sale restrictions, i.e., higher stock lending supply, does not lead to an increase in the occurrence of extreme negative stock returns. Helmes et al. (2017) investigate the short sale ban in Australia in 2008/09. Consistent with the other studies, they don’t find any supporting effects on stock prices, either. Bailey and Zheng (2011) report that the stock market downturn during the financial crisis 2008/09 cannot be attributed to short selling and that, therefore, the regulatory measures against it were not justified.
Even though not quite related to short sale bans, Boehmer et al. (2008) provide valuable insight into the relationship between shorting activity and stock returns. They find that heavily shorted stocks underperform weakly shorted stocks. This underperformance is persistent over time and strongest for small stocks, indicating that informed short sellers are able to identify poorly performing stocks and to bet against them. They do not, however, push down prices artificially below the fundamental value which would justify a regulatory measure against short sales.
This section comprises the liquidity analysis. I describe the variables and how I construct them and provide corresponding summary statistics before I introduce the methodology and give an overview of the empirical results including a modified identification strategy to assess the robustness of my results. Following the majority of studies described in the previous section, I hypothesize that the short sale bans in 2020 decreased stock liquidity.
Variable Construction. I investigate the impact of short sale bans on stock liquidity. Plenty of variables have been used to proxy stock liquidity. In this paper, my main variable to measure liquidity is a version of the bid-ask spread. To verify my results, I use the Amihud Illiquidity ratio and the dollar trading volume as two additional measures. To compute the bid-ask spread, I follow Amihud and Mendelson (1986) who suggest the percentage bid-ask spread , which is, for a given stock, defined as
Abbildung in dieser Leseprobe nicht enthalten
where is the closing ask price on day t, the closing bid price on day t and the closing price on day t. The authors argue that this measure exhibits a negative correlation with other variables that are used to proxy liquidity such as trading volume or number of shareholders. The availability of daily bid-and ask prices on Datastream are another argument in favor of this measure.23 A stock is considered more liquid, the lower the spread is. In this paper, I use the expressions “bid-ask spread”, “percentage bid-ask spread” and “spread” synonymously. I include the illiquidity measure introduced by Amihud (2002) and the dollar trading volume to complement the investigation on the effects on liquidity. For a given stock, the Amihud illiquidity ratio is expressed as
Abbildung in dieser Leseprobe nicht enthalten
where is the absolute log-return on day t and the number of shares traded on day t. The higher the Amihud illiquidity ratio, the more illiquid the stock. A stock is considered less liquid if ceteris paribus the daily dollar trading volume is low or the daily relative price change is high in absolute terms. There is no intuitive and straightforward meaning, though one can interpret the Amihud illiquidity measure “as the daily price response associated with one dollar of trading volume, thus serving as a rough measure of price impact.”24 To account for the very small numbers produced by this measure, I multiply it by 1,000,000. The third liquidity proxy is the dollar trading volume, which can be expressed as
Abbildung in dieser Leseprobe nicht enthalten
For a given stock on day t, it is defined as the number of traded stocks multiplied by the closing stock price, thus serving as a measure of turnover. Brennan et al. (1998) state that “this variable is associated with liquidity”25 and Chordia et al. (2001) use the same measure to proxy liquidity in their paper. Note that, even though I use the term dollar volume, all my data is denoted in Euro currency. Furthermore, I compute a dummy variable to capture the effect of short sale bans. The dummy equals one if a given stock is banned from short selling on day t and zero otherwise:
Abbildung in dieser Leseprobe nicht enthalten
Data. I use data from 28 European countries (27 EU countries + UK). This is somewhat different from other papers as the regulatory framework in all these countries is the same. Naked short sales have been forbidden in the EU since 2012 anyway and the threshold above which a short seller must disclose the short position is the same for all countries. Six out of 28 countries introduced short sale bans. Since the ban lasted two months, the time span of the data reaches from January 17, 2020, until May 18, 2020 (hereinafter referred to as the estimation period). I shift this period by two months in a later specification to test for effects of the ban lift. I retrieve daily close prices, as well as daily bid- and ask prices and daily trading volumes from Datastream. Additionally, the data set includes market value data for each firm as of January 1, 2020. For each country I apply the following filters: I select only the main national stock exchange, the domestic currency, major shares, and primary shares. The former makes sure that I include only the main equity instrument whereas the latter ensures that no share is included which is listed on a European stock exchange although the primary listing is outside Europe For example, Apple is also traded on the Milan Stock Exchange but is not subject to the investigation. Furthermore, I exclude ETFs and other securities that are traded as equity on a stock exchange but are not the major equity instrument of a publicly traded company. To compute Euro trading volumes and to rank firms according to their market value, all data is downloaded from Datastream in Euro. I exclude stocks for which the time series of the computed variables does not cover at least 80% of the days. Following related literature, I drop the top- and bottom 1% of the spread distribution from the (sub-) sample(s) to account for outliers and negative spreads, which, by definition, do not exist and do not make sense economically. After the raw dataset has been cleaned, I am left with a total of 5,130 stocks and 377,958 daily observations of the bid-ask spread. I apply the same data cleaning procedure to the other two variables for which I present summary statistics later in this section. For the bid-ask spreads, in 15.71% of the total daily observations, the ban dummy equals 1 and zero in the remaining 84.29%, indicating that roughly one-sixth of the stock-day observations were subject to the ban. Recall that only six out of 28 countries in the sample introduced a ban, which explains this relatively low number. Table 1 provides summary statistics of the three variables.
Table 1 : Summary Statistics
This table provides descriptive statistics of the daily observations after the data adjustments described in the text of the three liquidity proxies (bid-ask spread, Amihud ratio, and dollar volume) during the period between January 17 and May 18. For practical reasons the Amihud ratio is multiplied by 1 million and the dollar volume is denoted in a thousand Euros. Although I use the term “Dollar Volume” as common in the literature, all values are denoted in Euro currency.
Abbildung in dieser Leseprobe nicht enthalten
[...]
1 See Statista (2020).
2 The diff-in-diff structure refers to the fact that I do not only want to test if stock liquidity changes due to a short sale ban but whether this change (or difference) is significantly different from the difference produced by the control sample.
3 See D’Avolio (2002), p. 275-276.
4 See Hull (2015), p. 50.
5 See Hull (2015), p. 214-215.
6 See Hull (2015), p. 213.
7 See Boehmer et al. (2008), p. 523.
8 See Boehmer et al. (2013), p. 1396.
9 See McCaffrey (2010), p. 483.
10 See Dechow et al. (2001), p. 81.
11 ETFs are financial instruments that represent an entire basket of securities such as major stock indices. They are tradable like normal stocks on stock exchanges. See Mohamad et al. (2016), p. 1.
12 See Petram (2011), p. 24-28.
13 See Macey et al. (1989), p. 801.
14 See SEC (2008).
15 The assumption is that depositors view sharply declining share prices as a signal of bad firm performance.
16 See Younglai (December 31, 2008).
17 While Austria, France, and Spain set a threshold of 50%, Belgium, Greece, and Italy set a threshold of 20%. From April 15 onwards, the threshold was set uniquely among all countries at 50%.
18 ADRs are certificates that represent a legal claim on shares of a firm outside the US and whose shares are not listed on a US-American stock exchange. See Jayaraman et al. (1993), p. 91.
19 See Beber and Pagano (2013), p. 344.
20 See Boehmer et al. (2008), p. 525.
21 Therefore, the variable is not binary (as in my paper) but continuous, i.e., a likelihood for short sale regulation.
22 This is contrary to the famous Capital Asset Pricing Model by Sharpe (1964), where all investors have homogeneous beliefs about the firm’s future cash flows.
23 Note that despite the reasonably good data coverage, I still drop 1,466 out of 6,596 firms because they do not have enough data available or are excluded as part of the data cleaning.
24 See Amihud (2002), p. 32.
25 See Brennan et al. (1998), p. 351.
Der GRIN Verlag hat sich seit 1998 auf die Veröffentlichung akademischer eBooks und Bücher spezialisiert. Der GRIN Verlag steht damit als erstes Unternehmen für User Generated Quality Content. Die Verlagsseiten GRIN.com, Hausarbeiten.de und Diplomarbeiten24 bieten für Hochschullehrer, Absolventen und Studenten die ideale Plattform, wissenschaftliche Texte wie Hausarbeiten, Referate, Bachelorarbeiten, Masterarbeiten, Diplomarbeiten, Dissertationen und wissenschaftliche Aufsätze einem breiten Publikum zu präsentieren.
Kostenfreie Veröffentlichung: Hausarbeit, Bachelorarbeit, Diplomarbeit, Dissertation, Masterarbeit, Interpretation oder Referat jetzt veröffentlichen!
Kommentare