The effect of the 2008 short selling ban on the
U.S. stock returns and market liquidity
Abstract
On the 19th of September 2008 the SEC implemented a ban on short selling on nearly 800 financial stocks. This ban lasted 14 trading days, until the 8th of October 2008. In this paper the effects of this ban on stock returns and market liquidity will be examined. The regressions show that both the stock returns and the bid-ask spread – a measure of market illiquidity – increased as a result of the ban.
Name: Luca Veenis Supervisor: Mr. Robin Döttling MSc
Student number: 11029641
Bachelor: Economie & Bedrijfskunde Track: Economie & Financiering
University of Amsterdam Faculty of Economics and Business
Statement of originality
This document is written by Student Luca Veenis who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Introduction
Short selling has been around since some time now. You can find the origin of trading with common stocks – so, also the beginning of short selling – in the Dutch Republic of the early 1600s, where Isaac Le Maire acted as one of the first who deliberately tried to bring the price of particular – in this case, VOC – stocks down (van Dillen, 1930). Since then, there is not an unanimous thought about short selling, because some use it as a way to blacken others, while others just try to profit from depressions. This is why short selling is often accompanied by bans. These bans are usually implemented in times where it is not going well economically, just like the financial crisis which started in 2007. In September 2008, the SEC was forced to take action, because it was concerned that short selling of financial stocks was market manipulation which threatens investors and capital markets (U.S. Securities and Exchange Commission, 2008). Given the confidence the SEC has in the U.S. financial markets, they became concerned about the unexpected decreases in the prices of a lot of financial securities. These decreases can harm the value of stocks even more, because when people know that a lot of financial stocks are being used for short selling, the growth expectation of these stocks decreases as well. Given this decline in expectation, the value of stocks decreases even more. This is why the SEC temporary banned short selling on nearly 800 financial stocks. Initially, this ban would last for 10 business days, but it could be extended by no more than 30 business days. Eventually the ban lasted from the 19th of September 2008 until the 8th of October, so it was extended for a couple of business days.
In this paper, the effects of this ban on stock returns and market liquidity will be examined. The market liquidity will be measured in terms of the bid-ask spread. All of the firms that are on the S&P500 will be included in the regression and the distinction between the firms that got banned and the firms that did not get banned will be made by introducing a dummy variable where 1 equals banned and 0 equals not banned. In this way, a coefficient will be generated that is going to show to which extent the banned firms differ from the non-banned firms in terms of stock returns and market liquidity. The time period that is going to be used starts on the 19th of March 2008 and ends on the 8th of April 2009: six months prior to the
beginning of the ban and six months after the ban ended.
The data will be extracted from Wharton Research Data Services (WRDS). It will include the Closing Bid, Closing Ask and Holding Period Return. Given the Closing Bid and Ask prices, I will calculate the bid-ask spread. Given this data two regression will be performed. The stock returns are regressed using the Fama-French three-factor model plus two additional
dummy variables, and the market liquidity is regressed using an equation including various dummy variables. These regressions give the following results: the stock returns increased as a result of the short selling ban, while the market liquidity decreased.
First, this paper will define short selling and give its importance in the financial world. Then the literature that was used will be reviewed. Furthermore, the hypothesis and methodology will be displayed and clarified. In addition to this, the data will be summarized and regressed; the regression will be done in STATA. After all this, a conclusion will be given.
Definition
The most common form of trading with stocks is ‘buy low, sell high’: if one expects the price of a specific stock to go up, he buys the stock and profits when he sells it later at a higher price. This is called a long position. In contrary to a long position, an agent could also take a short position in a specific stock. This procedure is the exact opposite of what the agent does when he has a long position: he expects the price to go down, so instead of buying the stock, he is going to sell it with the intention of buying it back at a later date at a lower price (Marsh & Payne, 2012). Buying it back at a lower price means he made a profit (if the premium exceeds the fee he has to pay). Here is where the distinction between covered and naked short selling comes in place.
With covered short selling you first borrow the shares. In order to be able to borrow it, a fee must be paid to compensate the seller for the risk that the shares are not going to be returned. When the price is at a reasonable low, the shares get bought by the original buyer and returned to the original seller.
With naked short selling the trader sells shares that he does not own, assuming that the shares can be acquired in time to meet any delivery date (Bodie, Kane & Marcus, 2014). Using this way of short selling, no fee has to be paid, because the shares are not borrowed from a broker.
Literature review
In the past few years, extensive research has been done about the effect of short selling bans on for example stock prices, and a lot of opinions have been given. This research was performed in Australia, Hong Kong, the United Kingdom and the United States to name a few. Helmes, Henker, J. & Henker, T. (2017) examined the effects of the short selling ban on trading with financial stocks in Australia. They used Australian financial stocks as the treatment group and compared these stocks to the control group that includes Canadian financial stocks listed on the Toronto Stock Exchange. The effect was measured by doing univariate and multivariate fixed effects panel regressions. The initial goal of the Australian Securities and Investments Commission (ASIC) was to counter the negative effects on the stock prices that the short sellers caused, and thus raise the price. This goal was reached, but Helmes, Henker, J. & Henker, T. (2017) discussed that the ban created more costs than benefits. These costs include the following: there is enough evidence to conclude that the market quality – measured in trading activity, bid-ask spread and intraday volatility – of the stocks subject to the ban decreased.
The market quality in the United States suffered as well, state Boehmer, Jones & Zhang (2013). They found that the market quality – measured in quoted and effected spreads, price impacts and realized spreads – worsens more for big cap stocks1 than small cap stocks. However, stocks that already had some short selling restrictions were less affected than the ones without previous restrictions. Also, the price of the banned stocks rose, but they think this is because traders had the bailout program in mind. Just like the ASIC, the SEC wanted to (artificially) raise the prices, but Boehmer, Jones & Zhang (2013) argue whether the SEC should have taken the steps they took in order to reach this goal. The prices of the financial stocks rose indeed, but just like in Australia, this came with a cost. They question the advantages of this ban the market participants experience, i.e. how high should the benefits be in order to exceed the costs?
Tian (2012) also studied the effects of the temporary short selling ban on the financial stocks in the United States. He stated that previous research had to use non-financial stocks as a control group, because almost all of the financial stocks were banned. As a solution, he created a synthetic portfolio of stocks that did not get banned and used this as the control group. He found that the stocks were overvalued and that this overvaluation could be divided into two
parts. The first part explains the upward bias caused by the belief difference2 and the second
part is caused by speculative behaviour. The first part will slowly converge to zero, because when the ban is close to the end, there is not much time left to collect dividends. The second part will converge to zero too, because the time left to sell your stock to another party is also getting smaller. Tian (2012) proved this by using the Scheinkman and Xiong (2003) model in a finite trading horizon with no costs of trading.
The Hong Kong stock market also experienced stock overvaluation, conclude Chang, Cheng & Yu (2007). They came to this conclusion by identifying certain events in which short selling bans were applied to individual stocks. They also say that the overvaluation of the stocks is consistent with the findings of Miller (1977), who states that warrants permitted from short selling are likely to overpriced. The overvaluation of the stocks is higher when the spread of investor opinions is larger (Chang, Cheng & Yu, 2007). Next to the stock overvaluation, they found that no restrictions on short selling lead to high volatility for individual stock returns and less positive skewness.
In addition to the findings of Chang, Cheng & Yu (2007), Diether, Lee & Werner (2009) also found that the volatility of stocks is higher when short selling is allowed. They discovered that there is a positive relationship between short selling and simultaneous volatility. They used data from companies whose stocks are listed on the NYSE and Nasdaq. Their final conclusion is that the short sellers are not as bad as the media say they are. Their evidence shows that the short sellers, just like normal traders, help to correct short-term overreaction of stock prices to information.
Unlike the two abovementioned papers, Charoenrook & Daouk (2009) discovered that short selling is not associated with higher volatility. They used two kinds of data in their research: data on short selling regulation and practicality from 111 countries, and data on put option trading over the history of these countries. Their research focuses of examining the effect of the existence of short selling on the market returns, level of skewness and liquidity. They find, when short selling is allowed, that the accumulated returns are less volatile and that the market liquidity is of a higher level than when short selling is not allowed or has restrictions. However, they could not find evidence or potential reasons that restrictions on short selling have an effect on the level of skewness.
2 A stochastic differential equation (SDE) based on the proposition 1 in Scheinkman and Xiong (2003), which shows the difference in investors’ belief – i.e. how good an investor thinks his information is – between two groups.
Marsh & Payne (2012) also examined the effect of a short selling ban. They examined the effects of such a ban on the microstructure and the market quality in the United Kingdom. This effect was introduced in September 2008 and lifted in January 2009; they used data from June 2008 until February 2009 to cover the periods without the ban and the period with the ban. Just like Charoenrook & Daouk (2009), they find that the market liquidity of financial stocks decreased significantly. Also, they find that the costs of trading increased. Because of these increased costs, the volume of trades decreased. During the period the ban was active, the markets of financial stocks were far less efficient, and the part the trading process normally plays in stimulating price discovery was not as big as usual. When the ban was not active anymore, a big part of the above effects on market quality was reversed.
One could also examine the effects of short selling constraints on the adjustment speed of security prices, and this is exactly what Diamond & Verrecchia (1987) did. They try to do this by using a simple rational expectations model with bid and ask prices. This model gives outcomes on how security prices are distributed, and on the absolute and relative adjustment speed of prices to private information. They state that constraints on short selling do not have an effect on the biasedness of prices if the agents are rational. Also, short selling constraints have an influence on the proportion of private information that is exposed to the traders through observable trading. The effect short selling constraints have on informational efficiency has two parts. In the first part, for both uninformed and informed traders short selling is forbidden, and in the second part, only the uninformed party is restricted. They find that informational efficiency reduces in the first part but gets enhanced in the second part. In the overall effect, however, the reduction is stronger than the enhancement.
Instead of focusing your research on one country, one could also expand his research to a global scale. Beber & Pagano (2013) examined the effects of short selling bans in over 100 countries. The main focus of their study is the market liquidity of these countries, but price discovery and stock prices itself for example are also mentioned and studied. With regard to market liquidity: the bid-ask spreads increased significantly as a result of short selling bans. They also find that bans on short selling in the home market affects both the bid-ask spreads of the home and foreign market. In both of these market, the bid-ask spread increases. Short selling bans in the foreign market, however, only decrease the market liquidity of the foreign market. Another finding of Beber & Pagano (2013) is about price discovery: short selling bans have a negative effect of price discovery, because these bans slow this down. Their conclusion about
the stock prices is as follows: apart from the United States, the short selling bans did not make sure that the stock prices increased.
The impact of the short selling ban on equity option markets: this is what Battalio & Schultz (2011) examined. They address four questions: the first question focuses on the usage of the options market to avoid the short selling ban, the second focuses on the bid-ask spreads of options, the third focuses on the probable existence of biases in the relative prices of stocks and options, and the final one addresses whether or not price discrepancies emerged because of the short selling ban. They find that investors did not go to the option market to evade the short selling ban. Next to that, the bid-ask spreads increased for options on banned stocks. Also, synthetic share prices of the banned stocks were less than the real share prices. At last they find that the short selling ban had a positive effect on the amount of arbitrage opportunities.
The existing literature looked at for example the stock prices itself, bid-ask spread or (intraday) volatility. I will examine the stock returns – measured in holding period return – and bid-ask spread. In contrast to the reviewed literature I will use an existing model – Fama-French three-factor model – to regress my stock returns. This implies that my research adds another insight to the literature, namely the effect of the short selling ban on the holding period returns instead of the effect on the stock prices itself. My regressions will also show the difference in bid-ask spread and stock returns between the periods before, during and after the ban using dummy variables. My paper adds to the existing literature by examining the bid-ask spread (and thus, the market liquidity). The results I get are in line with the existing literature. An in-depth explanation of this is discussed in the data-analysis.
Hypotheses
Given the intention of the SEC, i.e. counter the negative effect short selling has on financial stocks, I would expect the prices of the banned firms to go up. The reviewed literature, however, suggests otherwise for specific countries. Helmes, Henker, J. & Henker, T. (2017) found that ASIC in Australia succeeded in raising the prices, while Beber & Pagano (2013) concluded that most countries did not succeed in raising the prices by introducing a ban on short selling. This applied to all but one country: the United States. Since Boehmer, Jones & Zhang (2013) also concluded that the prices in the U.S. rose, my first hypothesis will be:
The 2008 short selling ban in the U.S. will have a positive effect on the returns of the financial stocks.
Given the definition of market liquidity and the probable effect of a short selling ban, I would expect the market liquidity to decrease as a result of the ban. This is something all of the researchers (who examined the effect of a short selling ban on market liquidity) acknowledge, so my second hypothesis will be:
The 2008 short selling ban in the U.S. will have a negative effect on the market liquidity, and thus a positive effect on the bid-ask spread.
Methodology
To examine the effect of the short selling ban on the stock returns, the Fama-French three-factor model will be used. This is an asset pricing model that is an expansion of the original Capital Asset Pricing Model (CAPM) designed by Sharpe and Lintner. The CAPM formula is as follows: 𝐸[𝑅𝑖𝑡] − 𝑅𝑓𝑡 = 𝛼𝑖+ 𝛽𝑖𝑀(𝐸[𝑅𝑀𝑡] − 𝑅𝑓𝑡) + 𝜀𝑖𝑡. The Fama-French three-factor model
expands this model by adding size and value risk factors; these are given by SMB (Small Minus Big) and HML (High Minus Low). The formula now looks like this:
𝐸[𝑅𝑖𝑡] − 𝑅𝑓𝑡 = 𝛼𝑖 + 𝛽𝑖𝑀(𝐸[𝑅𝑀𝑡] − 𝑅𝑓𝑡) + 𝛽𝑖𝑠𝐸[𝑆𝑀𝐵𝑡] + 𝛽𝑖ℎ[𝐻𝑀𝐿𝑡]. With small minus big the difference between the returns on diversified portfolios of small and big stocks is meant and with high minus low the difference between the returns on diversified portfolios of high and low book-to-market stocks are meant (Fama & French, 2004). In the regression the holding period return will be the dependent variable and the risk-free rate, market risk-premium, size factor and value factor will be the independent variables. The holding period is calculated as follows: 𝐻𝑃𝑅 =𝐸𝑛𝑑𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑒 𝑜𝑓 𝑎 𝑠ℎ𝑎𝑟𝑒−𝐵𝑒𝑔𝑖𝑛𝑛𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑒 𝑜𝑓 𝑎 𝑠ℎ𝑎𝑟𝑒+𝐶𝑎𝑠ℎ 𝑑𝑖𝑣𝑖𝑑𝑒𝑛𝑑
𝐵𝑒𝑔𝑖𝑛𝑛𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑒 . I will add a
dummy variable which equals 1 when the firm is banned and 0 when the firm is not banned, so this is an additional independent variable (the dummy will only be 1 in the period of the ban). The final dummy variable that I will add equals 1 in the period during the ban and 0 otherwise. The equation that will be used for the regression looks like this: 𝐸[𝑅𝑖𝑡] = 𝛼𝑖+ 𝛽1(𝐸[𝑅𝑀𝑡] − 𝑅𝑓𝑡) + 𝛽2𝐸[𝑆𝑀𝐵𝑡] + 𝛽3[𝐻𝑀𝐿𝑡] + 𝛽4𝑅𝑓𝑡 + 𝛽5𝐵𝑎𝑛 + 𝛽6𝐵𝑎𝑛𝑃𝑒𝑟𝑖𝑜𝑑
To examine the effect of the short selling ban on the market liquidity, the bid-ask spread will be used. The bid-ask spread is calculated as follows: 𝐴𝑠𝑘 𝑝𝑟𝑖𝑐𝑒−𝐵𝑖𝑑 𝑝𝑟𝑖𝑐𝑒
𝐴𝑠𝑘 𝑝𝑟𝑖𝑐𝑒 ∗ 100%. The
bid-ask spread will not be included into a specific model, but it will be regressed using the following equation: 𝑌𝑖𝑦 = 𝛽1𝐵𝑎𝑛 + 𝛽2𝑃𝑟𝑒𝐵𝑎𝑛𝑃𝑒𝑟𝑖𝑜𝑑 + 𝛽3𝐵𝑎𝑛𝑃𝑒𝑟𝑖𝑜𝑑 + 𝐶 where the dependent variable is the bid-ask spread. Ban is a dummy variable that equals 1 if the firm got banned from short selling and 0 otherwise, PreBanPeriod is a dummy variable that equals 1 if the date is a date before the ban and 0 otherwise, and BanPeriod is a dummy variable that equals 1 if the date is a date in the short selling ben period and 0 otherwise. Finally, the C is a constant. Like in the Fama-French three-factor model, the Ban-dummy will only be 1 in the short selling ban period itself. A control variable will not be added, because this will not change the effect the short selling ban has on the bid-ask spread. This control variable will have an effect on the coefficient (it may increase or decrease), but this is not what this paper is about. The dummies
PreBanPeriod and BanPeriod are added to this equation to distinguish the three periods my data includes – the period before the ban, during the ban, and after the ban.
Data-analysis
The sample that is used in this paper consists of the firms active in the S&P 500 in the following period: six months prior to the implementation of the short selling ban until six months after the ban ended. The period that I will use consist of 267 trading days; the ban itself lasted for 14 trading days. The S&P 500 is an American stock market index which includes 500 large companies whose common stock is listed on the NYSE or the NASDAQ. This index is computed by calculating the total market value of the 500 companies of the current trading day versus the market value of the previous trading day (Bodie, Kane & Marcus, 2014). From the nearly 800 companies the SEC banned, only 59 were active on the S&P 500 in the sample period, so I will use these 59 firms as the experimental group. The other firms that were active on the S&P 500 act as the control group. The data that will be extracted from Wharton Research Data Services (WRDS) consists of the Closing Bid, Closing Ask and Holding Period Return. Given the Closing Bid and Ask prices, I will calculate the bid-ask spread.
The following pages describe the summary of the two regressions. Both are pooled OLS regressions. The first regression that is described is the one with regards to the stock returns. This description includes the betas of the dummy variables and other variables. From the betas a conclusion can be drawn about the effect of these variables on the stock returns.
The section after that describes regression with regards to the bid-ask spread. This description includes the betas of the specific dummy variables. From these betas a conclusion can be drawn about the effect of the dummy variables on the bid-ask spread.
All of the tables described above are displayed in the Appendix. The graphs that show the trend of the average holding period return and bid-ask spread are also shown here.
In the first regression the stock returns are regressed using the Fama-French three-factor model plus two dummy variables. The coefficient of the dummy variable Ban is positive, which implies that when the firm was banned, their return in the ban period is higher relative to the returns of all the firms in the non-ban period (keeping all other variables equal). This coefficient equals 0.0000676 which converts to 0.00676%; this shows that the short selling ban increased the returns of the banned firms by a very small positive number. Examining the other variables gives an insight in the way the returns are being influenced by those factors. Both the size and risk factor have a negative effect on the stock return, i.e. they lower the stock returns by respectively -0.0001146 (= -0.01146%) and -0.0026535 (= -0.26535%). The R-squared of this regression equals 0.014, which means that only 1.4% of the variable variation is explained by this model. A possible reason for this is the fact that the stock returns need additional control variables to explain more of the variable variation.
Given the effect of the short selling ban on the stock returns – the short selling ban increased the returns of the banned firms, my conclusion does indeed correspond with the existing literature. From an economic perspective it can be concluded that the short selling ban made sure that no investors could bet on the downfall of the banned firms anymore, and thus could not bring the expectations of the stock returns down.
The returns itself are also summarized. First, the returns are summarized for the ones where the dummy equals 1, i.e. when the firm is banned from short selling. Second, the returns are summarized for the ones where the dummy equals 0, i.e. all the firms in the non-ban period. Remember that the dummy variable is set up in such a way that it only equals 1 in the ban period. Analysing these results gives an average return of -0.015874 (= -1.5874%) for the banned firms in the ban period, while the average return of the remaining companies in the non-ban period equals -0.0008625 (= -0.08625%). The average return thus decreased with 0.0150115 (= 1.5015%) for the banned firms in the ban period, while the regression pointed out that the short selling ban had a positive effect on the returns. What also can be concluded is the fact that the spread of the returns decreased for the banned firms, but that is probably the case because less firms are banned from short selling relative to the firms that did not get banned (a lot less observations mean that the probability of a very high/low return is less than when there are a lot of observations). The standard deviation, however, increased for the banned firms relative to the non-banned firms.
In the second regression the bid-ask spread is regressed. The coefficient of the dummy variable Ban is positive, which implies that when the firm was banned, their bid-ask spread in the ban period is higher relative to the bid-ask spread of all the firms in the non-ban period (keeping all other variables equal). This coefficient equals 0.0026988, which shows that the bid-ask spread of the banned firms in the ban period is 0.26988% higher than the bid-ask spread for all the firms in the non-ban period. This means that the firms that were banned from short selling experienced a decrease in the market liquidity. Also, PreBanPeriod has a negative coefficient of -0.0020693, which means that the period prior to the ban experienced a lower bid-ask spread of 0.020693% relative to the period after the ban. In contrast to PreBanPeriod, BanPeriod has a positive coefficient of 0.0006103, which means that the ban period experienced a higher bid-ask spread of 0.06103% relative to the period after the ban. These conclusions can be drawn if assumed that the other variables are being kept equal. The R-squared of this regression equals 0.0014, which means that only 0.14% of the variable variation is explained by this model. A reason for this is the fact that the bid-ask spread needs additional control variables to explain more of the variable variation. As described in the methodology, I deliberately did not include a control variable so I expected a low R-squared. But again, a control variable would not change the effect of the short selling ban on the bid-ask spread, only the coefficient itself.
Given the effect of the short selling ban on the bid-ask spread – the short selling ban increased the bid-ask spread of the banned firms, my conclusion does indeed correspond with the existing literature. From an economic perspective it can be concluded that the short selling ban made sure that the market became less liquid, because the ban prohibited market participants to provide liquidity to the market. Short selling was a good way to provide the market with a great deal of liquidity, and this is way the short selling ban enhanced the effect on the market liquidity.
The bid-ask spread itself is also summarized. First, the bid-ask spread is summarized for the ones where the dummy Ban equals 1, i.e. the bid-ask spread for the banned firms in the ban period is summarized. Second, the bid-ask spread is summarized for the ones where the dummy Ban equals 0, i.e. the bid-ask spread for all the firms in the period before and after the ban is summarized. The average bid-ask spread for the banned firms in the ban period equals 0.006831 (= 0.6831%), while the average bid-ask spread in the period before and after the ban equals 0.0025942 (=0.25942%). This shows that the average bid-ask spread for the banned firms increased by 0.0042368 (= 0.42368%). What also can be concluded from these tables is the fact that the difference in maximum and minimum bid-ask spread is lower for the banned
firms in the ban period than for the firms in the non-ban period. But again, this is probably the case because the observations for the banned firms in the ban period are much lower than the observations for the firms in the non-ban period. The bid-ask spread, however, decreased for the banned firms relative to the non-banned firms.
Conclusion
This paper examined the effect of the 2008 short selling ban on the stock returns and market liquidity in the United States. The data used to perform a regression on comes from the S&P 500. The time period used starts six months prior to the start of the ban and ends six months after the ban ended. In this period, a total of 545 firms were active on the S&P 500, but only 59 of them were banned from short selling, so these 59 firms act as the experimental group. The remaining firms on the S&P 500 act as the control group. To examine the effects of the short selling ban I used two hypotheses. The first hypothesis focuses on the stock returns, and an adjusted Fama-French three-factor model is used to regress this. It is adjusted because two dummy variables are added which tells whether the firm is banned or not and whether the date is inside the ban period or not. The second hypothesis focuses on the bid-ask spread – the measure used for market liquidity, and to regress this an equation with three dummy variables is used.
In table 1 (see Appendix for the tables) the first regression is summarized. From this table it can be concluded that the ban had a positive effect on the stock returns. This conclusion is in line with the reviewed literature, and thus the first hypothesis. In tables 1.a and 1.b the variable holding period return is summarized for the banned firms and non-banned firms respectively. Here can be seen that the mean of the holding period return is lower for the banned firms in the ban period in comparison with the holding period return in the non-ban period. Also, the standard deviation of the holding period returns is higher for the banned firms in the ban period relative to the non-ban period.
In table 2 the second regression is summarized. From this table it can be concluded that the short selling ban h ad a positive effect on the bid-ask spread, and thus a negative effect on the market liquidity. This conclusion is also in line with the reviewed literature, and thus my second hypothesis. In tables 2.a and 2.b the variable bid-ask spread is summarized. Here can be seen that the mean of the bid-ask spread for the banned firms in the ban period is higher than the bid-ask spread in the non-ban period. Also, the standard deviation of the bid-ask spread is lower for the banned firms in the ban period relative to the non-ban period.
If this topic is going to be researched further, one can look at the effect of the short selling ban across multiple countries. Also, in this paper the control group consisted of mostly non-financial stocks, while the best comparison with the banned financial stocks would be a control group that also consists of financial stocks that did not get banned. This, however, is
researched by Tian (2012). What also could be researched is the effect of the short selling ban on (intraday) volatility or the trading volume in the U.S. stock market. Another change one could make is expanding both regressions by adding various control variables to improve the quality of these regressions.
Bibliography
Battalio, R., & Schultz, P. (2011, December). Regulatory Uncertainty and Market Liquidity: The 2008 Short Sale Ban's Impact on Equity Option Markets. Journal of Finance, 66(6), 2013-2053.
Beber, A., & Pagano, M. (2013, February). Short-Selling Bans around the World: Evidence from the 2007-2009 Crisis. The Journal of Finance, 68(1), 343-381.
Bodie, Z., Kane, A., & Marcus, A. J. (2014). Investments - Global Edition (10th ed.). London: McGraw Hill Higher Education.
Boehmer, E., Jones, C. M., & Zhang, X. (2013). Shackling short sellers: The 2008 shorting ban.
Review of Financial Studies, 26(6), 1363-1400.
Chang, E. C., Cheng, J. W., & Yu, Y. (2007). Short-sales constraints and price discovery: Evidence from the Hong Kong market. The Journal of Finance, 62(5), 2097-2121. Charoenrook, A., & Daouk, H. (2009). A study of market-wide short-selling restrictions.
Vanderbilt University and Cornell University working paper.
Cox, C. (n.d.). EMERGENCY ORDER PURSUANT TO SECTION 12(k)(2) OF THE SECURITIES.
Diamond, D. W., & Verrecchia, R. E. (1987). Constraints on short-selling and asset price adjustment to private information. Journal of Financial Economics, 18(2), 277-311. Diether, K. B., Lee, K. H., & Werner, I. M. (2008). Short-sale strategies and return
predictability. The Review of Financial Studies, 22(2), 575-607.
Engelberg, J. E., Ringgenberg, M. C., & Reed, A. V. (2018, April). Short-Selling Risk. Journal
of Finance, 73(2), 755-786.
Fama, E. F., & French, K. R. (2004). The capital asset pricing model: Theory and evidence.
Journal of economic perspectives, 18(3), 25-46.
Helmes, U., Henker, J., & Henker, T. (2017). Effect of the ban on short selling on market prices and volatility. Accounting & Finance, 57(3), 727-757.
Marsh, I. W., & Payne, R. (2012). Banning short sales and market quality: The UK's experience.
Journal of Banking & Finance, 36(7), 1975-1986.
Miller, E. M. (1977). Risk, uncertainty, and divergence of opinion. The Journal of Finance,
Scheinkman, J. A., & Xiong, W. (2003). Overconfidence and speculative bubbles. Journal of
Political Economy, 111(6), 1183-1220.
Tian, H. (2012). The effects of a temporary short-selling ban. preprint.
van Dillen, J. G. (1930). Isaac Le Maire en de handel in actiën der Oost-Indische Compagnie. In M. Nijhoff, Economisch Historisch Jaarboek 16 (pp. 1-165). 's Gravenhage: De Vereeniging.
Appendix Table 1 Table 1.a Table 1.b _cons -.0003142 .0001932 -1.63 0.104 -.0006928 .0000645 BanPeriod -.0150242 .0006338 -23.70 0.000 -.0162664 -.0137819 Ban .0000676 .0018381 0.04 0.971 -.0035349 .0036702 Rf .065027 .0267997 2.43 0.015 .0125001 .1175539 HML -.0026535 .0001671 -15.88 0.000 -.002981 -.0023259 SMB -.0001146 .0001649 -0.69 0.487 -.0004379 .0002087 MktRf .0020077 .0000597 33.61 0.000 .0018907 .0021248 ret Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 335.735228 136,403 .002461348 Root MSE = .04926 Adj R-squared = 0.0140 Residual 331.028274 136,397 .002426947 R-squared = 0.0140 Model 4.70695398 6 .78449233 Prob > F = 0.0000 F(6, 136397) = 323.24 Source SS df MS Number of obs = 136,404
ret 806 -.015874 .1138065 -.9050888 .7219999 Variable Obs Mean Std. Dev. Min Max
ret 135,598 -.0008625 .0489669 -.6430657 1.452727 Variable Obs Mean Std. Dev. Min Max
Table 2 Table 2.a Table 2.b _cons .0035685 .0001185 30.11 0.000 .0033362 .0038008 BanPeriod .0006103 .0003918 1.56 0.119 -.0001576 .0013782 PreBanPeriod -.0020693 .0001662 -12.45 0.000 -.002395 -.0017435 Ban .0026988 .0011141 2.42 0.015 .0005152 .0048824 BidAskSpread Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 121.822962 136,399 .000893137 Root MSE = .02986 Adj R-squared = 0.0014 Residual 121.653391 136,396 .000891913 R-squared = 0.0014 Model .169570153 3 .056523384 Prob > F = 0.0000 F(3, 136396) = 63.37 Source SS df MS Number of obs = 136,400
BidAskSpread 806 .006831 .0171084 -.0147783 .25 Variable Obs Mean Std. Dev. Min Max
BidAskSpread 135,594 .0025942 .0299433 -.092559 8.400596 Variable Obs Mean Std. Dev. Min Max
Figure 3
In Figure 3 the average return of all the firms is shown, while in Figure 4 the average return for the banned firms on the S&P 500 is shown. The two red lines are respectively the beginning and end of the short selling ban. In figure 4 you cannot clearly see that the holding period returns decreased, but what can be seen is the expansion of the holding period return spread.
Figure 5
In Figure 5 the average ask spread for all firms is shown, while in Figure 6 the average bid-ask spread for the banned firms on the S&P 500 is shown. The two red lines are respectively the beginning and end of the short selling ban. In figure 6 you can clearly see the increase of the average bid-ask spread in the ban period.
