The impact of short selling ban imposed during the
2008-2009 crisis on market quality in the United
Kingdom
Abstract On Thursday the 18th of September 2008, only a few days after the fall of the Lehman Brothers, the Financial Service Authority (FSA) placed a ban on short selling on financial stocks. The ban was implemented to protect the quality of the market and to guard against further instability. In this paper, the market liquidity, measured in the bid ask spread, and price volatility are tested. If the effects of the short selling ban is weaker for Globally Systemic Important Banks is tested as well. A regression analysis is used to examine the effects. Evidence is found that the price volatility and the bid ask spread increase more for the banned firms than for the non-banned firms, due to the short selling ban. The regression results about those effects on GSIB’s point out that these are weaker. Name: Chris Keizer Name supervisor: Ieva Sakalauskaite Bachelor: Economics and Business Track: Economics and Finance1. Introduction On Thursday 18th September 2008, the Financial Service Authority (FSA) of the United Kingdom placed a temporary ban on short selling in 29 financial stocks from the London Stock Exchange. Within one month 6 stocks were added to that list and one was removed. On that same day, after the closing of the market, the Securities and Exchange Comission (SEC) matched the FSA by placing a ban on short selling on 797 financial stocks. The duration of these short selling restrictions would cover 10 business days with a possibility to extend the order duration by 30 calendar days (Boehmer, Jones and Zhang). The announcement of the ban came only a few days after The Lehman Brothers had fallen. The reason for the FSA to take this action, formulated by the FSA CEO Hector Sands, was to protect the integrity and quality of the market, and besides that to guard against further instability. In the evening after the announcement, FSA chairman Callum McCarthy, clarified further the ban and expressed he was concerned about the incoherence in trading of equities and the volatility. According to Callum McCarthy this short selling measure should have a calming effect (Hansson and Fors). The short selling ban was announced on Thursday 18th of September 2008 and implemented on the 19th of September. The ban ended the 16th of January 2009.
This paper will examine the effects of a short selling ban on the price volatility and liquidity, measured in the bid ask spread, of the affected stocks. The stocks of the banned firms that were banned by the FSA will be compared to the control group of all firms on the FTSE350 but not on the short selling list. The time period that is used, starts at the 1st of January 2008 and ends at the 31st of
December 2009. Firms that were not listed on the FTSE350 for the whole period of time were excluded from the sample. This paper is divided into 7 sections. It starts with the background information on short selling and Globally Systemic Important Banks, followed by a review of the relevant literature about this subject. After this, the hypothesis and methodology are shown and explained. The data that is used is presented in the section after this one, followed by the results that are obtained by doing a regression analysis. This paper finishes with the conclusion.
2. Background Information 2.1 Short Selling Agents sell securities short when they believe the stock is overvalued. They speculate that the price will go down and make a profit from that (Marsh & Payne). There are two basic types of short selling: ‘covered’ short selling and ‘naked’ short selling. Covered short selling means that a short seller borrows shares so that at the settlement the shares can be delivered to the buyer. The borrower pays the lender a fee in exchange for borrowing the stock. The short seller believes the price of the share they sell short will fall in the future, so at some point in the future the short seller buys the same amount of stock back and return the stocks to the original lender (Bodie, Kane and Marcus). Naked short sellers sell shares they, different from covered short selling, do not own. Because they don’t borrow the shares, a naked short seller does not have to pay borrowing fees. At the time the short seller wants to close the shorting position, he/she has to buy the same amount of shares he/she sold. This kind of short selling caries a settlement risk of not being able to deliver the stocks to the buyer at the time the settlement period ends (FSA). 2.2 Globally Systemic Important Banks Globally Systamic Important Banks, GSIB’s, are large banks which are considered to be “Too Big to Fail” (Laeven and Valencia). The Basel Committee on banking supervision decided to better regulate 29 large banks. For future policy and research, it is an important issue if these banks are better supervised and monitored (Moshirian). The Globally Systemic Important Banks in the United Kingdom are: Barclays, The Royal Bank of Scotland, Standard Chartered, Lloyds Banking Group and HSBC.
3. Literature review Over the years, many studies have been implemented about the effects of a ban on short selling and many different allegations about those effects have been made. Evidence on the effects of short selling restrictions is mixed. Miller (1977) suggests that the supply of stocks on the market will be lower when a ban on short selling is implemented. This is because with a possibility to sell short, the supply increases with the amount of stocks you go short with. Miller (1977) also states that stock prices will be overvalued when a short-sales constraint is implemented, because only optimistic investors will trade. This is, because pessimistic investors are not allowed to go short on stocks they believe the price will go down for. This results that only optimistic information is included in the stock price. Chang, Cheng and Yu (2007) found that short sale restrictions tend to cause stock overvaluation, which is consistent with Millers findings. They came to this finding by analyzing the Hong Kong stock market. They compared stock returns after they were added to the official short selling list and before they were on that list. This is a list where it is only possible to go short on securities if this security is on this list. Other results from this research suggest that in Hong Kong the returns of individual stocks are less positively skewed and the volatility of individual stocks is higher when short selling is practiced. In contrast with the results from Chang, Cheng and Yu(2007), Charoenrook and Daouk (2005) find strong evidence that volatility of an individual stock is lower when it is possible to sell short. They have used regulations and practices of put option trading and short selling from 1969 to 2002 for 111 countries, and reported new data on these. They have researched as well what the effect of short-selling is on skewness, market returns and liquidity. Result of their research is that there is no effect of the restriction on skewness and market returns. Other results, however, show that a constraint of short selling reacts differently on individual stocks than on the market. At market level the lower expected returns dominate whereas at individual level the overpricing might dominate. The results of the effect on liquidity point out that when short selling is not allowed, the liquidity is reduced.
Diamond and Verrecchia (1987) have done research into the effect of a short selling constraint on the adjustment speed of private information on security prices. They split this adjustment speed up in the absolute speed of adjustment of private information and the relative speed. They find that a constraint on short selling led to a decline in adjustment speed, especially for negative information, on prices. The information efficiency is reduced when a constraint is active. Diamond and Verrecchia find that the bid-ask spread stays the same when both the informed and uninformed are subject to a constraint on short selling. They predicted that the liquidity therefore would decrease, because of the lower information efficiency. Boehmer, Jones and Zhang (2009) studied in their paper nearly 1000 financial stocks that were banned for short selling by the SEC in 2008. They examined the effect of a ban on short selling on market quality, stock price and shorting activity by using a matched sample, whereby a nonbanned stock is matched with a banned stock. Results from this study are that the shorting activity drops by a 77% in large cap stocks and is mostly unaffected in small cap stocks. The market quality, which is measured in effective and realized spreads, degrades by the ban. Diether, Lee and Wernern (2009) examined the effects of short selling on intraday volatility, bid ask spreads and returns. The sample they used are US stocks listed on the NYSE and Nasdaq. They found that the intraday volatility increased after a short selling ban is lifted. They found as well that short sellers are able to correct an overreaction on the short term and an insignificant effect on market liquidity due to a short selling ban. Marsh and Payne (2012) did research to effect of restriction on short selling on the market quality, efficiency, trading volume and price discovery. They also investigated the market conditions and what motivated the ban on short selling. Basis points bid ask spread are used as a measure for liquidity and the sum of shares that were bought or sold on a day for trading volume. Their analysis is
market conditions for financial stocks are not different than market conditions for the stock from the control group before the ban was implemented. They find that the market liquidity for financial stocks fell and the trading costs increased dramatically. This resulted in reduced market quality, a smaller price discovery and high price impacts. Another research into the effects of the short selling ban in the United Kingdom is done by Hansson and Fors (2009). They investigate the effects on stock returns, volatility, trading volume and the bid ask spread. The data they used, is obtained directly from the London Stock Exchange. The sample period that is used consists of 183 trading days. Because this has a relatively short preban period, they split the ban period up in two periods so the sample consists of 4 parts. They have done this to examine the consistency throughout the ban period. Hansson and Fors found that, just as Marsh and Payne, the volatility of the stocks increase and the bid ask spread widens when the ban on short selling was active. They found also strong evidence that the trading volume decreased. Just like Marsh and Payne, they found a significantly deterioration in market quality. Clifton and Michayluk (2010) have done research into the effects of a short selling ban on the trading activity, stock volatility and market liquidity, measured in bid ask spread. They analyzed the London Stock Exchange for a period of 160 whole trading days starting from the 23rd of July 2008 till the 12th of march 2009. The results of their research are that the bid ask spread widens and the trading volume decreased. Clifton and Michayluk did research in both return volatility and price volatility. The results are that the price volatility had a peek when the ban was implemented and the return volatility increased as well. Beber and Pagano (2013) used for their study data for 17040 stocks from 30 countries. They used this data to identify the effects of a ban on short selling on stock prices, market liquidity and price discovery. With this information Beber and Pagano could make a distinction between the effects of a ban on ‘naked’ short selling and ‘covered’ short selling. The findings of this research are that a ban on short selling led to high volatility and has been detrimental for the market liquidity, especially for stocks that have a small market capitalization. The short selling ban had as effect on the price discovery that is slowed and it
had no effect on the stock prices. They conclude that the main payoff of the experiment to ban short sales is the amount of evidence it delivered that this should not be done again in the next financial crisis.
4. Hypotheses The hypotheses that are tested in this paper are based on research done described in the literature that is used. Much research is done into the effects of a ban on short selling on stock return volatility. In this paper, the intraday stock price volatility is tested and the banned stocks are compared with the non-banned stocks. In research done by Clifton and Michayluk (2010) is found that the intraday volatility of stocks increases when short selling restrictions are implemented. Therefore, the first hypothesis will be: Hypothesis 1: The intraday price volatility for the banned stocks will be higher than the stocks that are not banned. Charoenrook and Daouk (2005) have done research into the effects of short selling in 111 countries and found that the liquidity of a stock decreases when it is not possible to sell short. Boehmer, Jones and Zhang (2009) looked at the effects of the short selling ban imposed by the SEC in 2008. They found as well that the liquidity of stocks, measured in effective and realized spreads, decreased. On the contrary, Diamond and Verrecchia (1987) looked at the adjustment speed of private information on security prices. They found that the adjustment speed would decrease when a ban is active and that when a short selling ban had effect on both informed and uninformed investors, the bid ask spread would stay the same. Diether, Werner and Lee (2009) find as well insignificant effect of short selling ban on market liquidity. The bid ask spread, as a measure for liquidity, is tested in this paper as well and therefore the second hypothesis will be: Hypothesis 2: The bid ask spread will widen more for the banned stocks than for non-banned stocks.
Globally Systemic Important Banks (GSIB’s) are financial institutions that are considered to be “Too Big To Fail”. During the recent crisis, a broad range of measures is taken to reignite economic growth by capital injections, asset purchases and guarantees (Laeven and Valencia, 2012). It is likely to say that the effects of a short selling ban are weaker for GSIB’s. Therefore, the third hypothesis is: Hypothesis 3: The effects of a short selling ban are weaker on GSIB’s than on other banned firms.
5. Methodology 5.1 Volatility In order to examine the first hypothesis, the measure for the intraday volatility that is used is as follows: Intraday Price Volatility = (!"#$% !"#$% !"#!!!"#$% !"#$% !"#)!"#$% !"#$%&' !"#$% In this equation, Daily Price High (Low) stands for the highest (lowest) price of a stock during a particular day. The difference between the daily high and low price is divided by the daily closing share price. In research done by Boehmer, Jones and Zhang (2009), the price volatility is calculated by the daily high price minus the daily low price and divided by the stock’s volume-weighted average trading price. Because Datastream couldn’t provide that data about the stock’s volume-weighted average trading price, this paper makes use of the daily closing stock price. To perform a regression analysis on the effect of a short selling ban on the volatility, the following regression equation is used: 𝑌!" = 𝛽!+ 𝛽!𝐿𝑖𝑠𝑡!"+ 𝛽!𝑃𝑟𝑒𝐵𝑎𝑛!"+ 𝛽!𝑃𝑟𝑒𝐵𝑎𝑛!" ∗ 𝐿𝑖𝑠𝑡!"+ 𝛽!𝐵𝑎𝑛!"+ 𝛽!𝐵𝑎𝑛!" ∗ 𝐿𝑖𝑠𝑡!"+ 𝛽! 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 + 𝜀!" The dependent variable 𝑌!" in this equation is the intraday price volatility of a stock for stock i on time t. The independent variables are on the other side of the equation and 𝛽! is a constant. The first independent variable is 𝐿𝑖𝑠𝑡!", which is a dummy variable. This equals 1 if the firm is a banned firm, so on the short selling ban list, and it equals 0 if it is not. The two dummy variables 𝑃𝑟𝑒𝐵𝑎𝑛!" and 𝐵𝑎𝑛!" equal to 1 if the time period is before the ban and during the ban, respectively. The interaction between the pre-ban period and banned stocks, and the ban period and the banned stocks is captured in the interaction variables 𝑃𝑟𝑒𝐵𝑎𝑛!"*𝐹𝑖𝑛!" and 𝐵𝑎𝑛!"*𝐹𝑖𝑛!", respectively. The control variables that are included in this regression equation are the natural logarithm of the turnover of stocks and the market value of the firms, measured by the share price multiplied
with the number of ordinary shares in issue. Natural logarithms are used, because that makes data less prone to outliers and prevent misleading data. 5.2 Bid Ask Spread For the second hypothesis, the effect of a ban on short selling on the bid ask spread is examined. The bid ask spread is used as a measure for the liquidity of a stock. The bid ask spread is formulated as follows: Bid Ask Spread = !"#$% !"#!!"#$% !"#!"#$% !"#$%&' !"#$% ×100% To test this second hypothesis, this paper makes use of the same regression equation as before for the volatility. 𝑌!" = 𝛽!+ 𝛽!𝐿𝑖𝑠𝑡!"+ 𝛽!𝑃𝑟𝑒𝐵𝑎𝑛!"+ 𝛽!𝑃𝑟𝑒𝐵𝑎𝑛!" ∗ 𝐿𝑖𝑠𝑡!"+ 𝛽!𝐵𝑎𝑛!"+ 𝛽!𝐵𝑎𝑛!" ∗ 𝐿𝑖𝑠𝑡!"+ 𝛽! 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 + 𝜀!" In this case the dependent variable, 𝑌!", stands for the bid ask spread. The independent variables used are the same as in the regression equation above. The difference is that with this equation only logarithmic turnover is used as a control variable. This is because only logarithmic turnover is a good explanatory variable. 5.3 Globally Systemic Important Banks For the last hypothesis, we examine if the effect of a short selling constraint is less strong for a globally systemic important bank than for other banned firms. The regression equation that is used, is different and as follows: 𝑌!" = 𝛽!+ 𝛽!𝐿𝑖𝑠𝑡!"+ 𝛽!𝐺𝑆𝐼𝐵!"+ 𝛽!𝑃𝑟𝑒𝐵𝑎𝑛!"+ 𝛽!𝑃𝑟𝑒𝐵𝑎𝑛!"∗ 𝐿𝑖𝑠𝑡!"
The dependent variable and control variables used to test the volatility and the bid ask spread are still the same as were used to test the first and second hypothesis.
6. Data The sample that is used in this paper consists of firms that were listed on the FTSE350 of the United Kingdom. The FTSE350 is a list of 350 firms with the largest market capitalization of the United Kingdom. The data that is used, is daily data extracted from Datastream. The data that is extracted is the stock price, the turnover by volume, the ask price, the bid price, the highest and lowest daily price, and the market value.
The time period used starts on the 1st of January 2008 and end on the 31st of December 2009. The ban on short selling was imposed on the 19th of September 2008 and ended on the 16th of January 2009. The period in which the ban was active consists of 83 trading days. The amount of trading days before the ban period and after the ban period consists of 182 trading days and 242 trading days, respectively. This makes the sample period a total 507 trading days. The sample consists of firms that are included in the FTSE350 for the whole period from January 2008 till December 2009. The amount of firms started with 358 firms. Eventually, 11 firms were excluded from the final sample. 8 of those firms are from the control group and were scrapped from the sample, because they went bankrupt on some moment in this period of time. 3 of the excluded firms are banned firms and they were excluded as well because they didn’t survive through the whole period of time. The list of banned stocks consisted of 34 financial stocks. In this sample 22 are included and 12 were excluded. The reason for this is, because 9 firms were not traded on the FTSE350 but on the FTSE Small Cap Index and 3 firms defaulted in the period of time used. Hereby, a possible survivorship bias is acknowledged but oddly acting data avoided. The globally systemic important banks in this sample are The Royal Bank of Scotland, Barclays, Lloyds Banking Group, HSBC and Standard Chartered.
When we take a look at the bid ask spreads in table 1a, b and c, we see that for the banned firms the average bid ask spread widens to nearly 300 percent of the value before the ban on short selling was implemented. Before the ban was implemented, the bid ask spread was lower for the banned firms than for the non-banned firms. During the ban period it was higher for the banned firms. This means that before the ban, the banned stocks were more liquid and during the ban less liquid. For the non-banned firms, the bid ask spread increases with less than 100 percent of the value before the ban was implemented. Table 1c shows for the GSIB’s an increase of the bid ask spread of 200 percent of the value before the ban was introduced. The bid ask spread for the GSIB’s are both before the ban and during the ban the lowest. This means that the stocks of the GSIB’s are the most liquid both before the ban and during the ban. The information on the volatility show that the average price volatility increases for both banned firms and non-banned firms. In contrast with the bid ask spread, the volatility of the GSIB’s increase the most and to the highest level. The volatility of the non-banned stocks is before the ban and during the ban the lowest. Table 1 In the tables below, the summary statistics are shown for the banned firms, non-banned firms and GSIB’s. Table 1a. Banned firms Pre-ban period Ban period LogMV 4,510 8.227982 1.498026 5.742426 11.60113 LogTurnover 4,366 8.291894 1.951486 2.388763 13.47473 VolSpread 4,510 .0404602 .0246191 0 .346291 BidAsk 4,510 .164841 .2016207 -3.846242 3.455285 Variable Obs Mean Std. Dev. Min Max
LogMV 2,038 7.915658 1.441333 5.028737 11.62483 LogTurnover 1,966 7.670281 2.183915 .1823216 13.20041 VolSpread 2,038 .0856878 .0543794 0 .7388889 BidAsk 2,038 .5898306 .832425 0 8.271266 Variable Obs Mean Std. Dev. Min Max
Table 1b. Non-Banned firms Pre-ban period Ban period Table 1c. Globally Systemic Important Banks Pre-ban period Ban period Figure 1 shows the daily average bid ask spread for the banned firms, non-banned firms and the Globally Systemic Important Banks. The information provided by the summary statistics are consistent with this figure. We can see LogMV 60,250 7.155806 1.260182 4.179451 11.71577 LogTurnover 58,144 7.062482 1.791645 -2.302585 13.72651 VolSpread 60,250 .0377316 .0258399 0 .5483803 BidAsk 60,250 .3227433 .758709 -9.84375 100 Variable Obs Mean Std. Dev. Min Max
LogMV 27,452 6.720573 1.352775 2.426571 11.56091 LogTurnover 26,474 6.994028 1.769356 -1.609438 13.11434 VolSpread 27,452 .0694948 .0507258 0 .9916193 BidAsk 27,452 .5563181 1.391063 -4.940711 103.0664 Variable Obs Mean Std. Dev. Min Max
LogMV 939 10.44484 .5657182 9.511057 11.60113 LogTurnover 909 10.42066 1.08317 7.906547 13.47473 VolSpread 939 .0452011 .0310766 0 .346291 BidAsk 939 .0820372 .1184579 0 3.063996 Variable Obs Mean Std. Dev. Min Max
LogMV 425 9.95446 .7800813 8.864721 11.62483 LogTurnover 410 10.15186 1.097768 6.889286 13.20041 VolSpread 425 .0964989 .0720524 0 .7388889 BidAsk 425 .2357233 .3152069 0 2.534432 Variable Obs Mean Std. Dev. Min Max
decreased for all firms, but now the bid ask spread is highest for the non-banned firms just like before the ban period. In Figure 2, the daily average price volatility is shown. This figure is in line with the information from the summary statistics as well. When we take a look at the date when the ban was implemented, we see that the daily average price volatility increases for all firms. The GSIB’s and other banned firms experienced the highest increase and is the highest for the GSIB’s. After the ban period, the GSIB’s again experienced a peak in price volatility, whereas the other banned firms experienced only a small increase. After this peak, the volatility gradually decreases. For the non-banned firms the same applies, but it started almost directly after the ban was implemented. Figure 1: Bid Ask Spread
The daily average bid ask spread are shown in this figure for the banned firms, the non-banned firms and the Globally Systemic Important Banks (GSIB’s) for the period from January 1st , 2008 to December 31st , 2009. Vertical black lines represent the start and the end of the ban period, respectively.
Figure 2: Volatility In this figure the daily average price volatility is shown. In this figure, the same applies as in figure 1.
7. Results 7.1 Bid Ask Spread The findings of the regression analysis confirms the hypothesis that the bid ask spread for banned firms widens more than for non-banned firms, Table 2 shows the regression analysis of the bid ask spread. From regression (1) to regression (2), a control variable is included and in this case the control variable is the logarithmic turnover. When we take a look at the second column, we see that after adding the logarithmic turnover to the regression, the variable List is changed from highly significant to not significant at all. When we take a look at the second regression column, we can therefore see that the variable List has no impact on the bid ask spread. This means that for banned firms the liquidity is not significantly higher than for non-banned firms. The variables Preban and Ban are highly significant. The coefficient of Preban is negative and the coefficient of Ban is positive. The bid ask spread is therefore lower in the preban period than in the ban period for both banned and non-banned firms. This means that the liquidity is higher in the period before the ban than during the bun. The interaction variable PrebanList is not significant, so there is no significant difference in bid ask spread, liquidity, between banned firms and non-banned firms in the pre-ban period. This is in line with Figure 1 where the bid ask spread increases for both banned and non-banned firms when the ban was implemented. The interaction variable BanList is as well highly significant. This means that effect of a short selling ban is negative for the liquidity. The effect of a ban is that the bid ask spread widens with 0.1436 points and therefore the liquidity reduces. This is in line with figure 1 as well. The bid ask spread increases when a ban on short selling is implemented. We can see that the firms that are subject to the ban have a higher bid ask spread. The control variable LnTurnover stands for the logarithmic turnover and this variable is highly significant. The dependent variable Bid Ask Spread has a linear-log relationship, which means that a 1 percent change in the turnover has an effect of 0.01 x βLnTurnover. In this case, this means that a firm with a 1
percent higher turnover has 0.01 x -0.1299 lower bid ask spread. The liquidity therefore increases when the turnover is higher. 7.2 Price Volatility The results from the regression analysis confirm the first hypothesis that the price volatility is higher for the banned firms than for the non-banned firms. Table 3 shows the regression analysis for the price volatility. From regression (1) to regression (2), the control variable logarithmic turnover is included and from regression (2) to regression (3), the control variable logarithmic market value is included. When we take a look at the third column, the variable List has a significant effect on the volatility with positive coefficient of 0.0099 if it is a banned firm. This means that on average the short selling ban increases the volatility. The variables PreBan and Ban have a high level of significance. For the variable PreBan, the coefficient is negative and for the variable Ban it is positive. The meaning of this is that when a firm is in the period before the ban, it has a lower average price volatility than when a firm is in the ban period. When a firm is a banned one and it is before the ban was implemented, it has, according to the highly significant interaction variable PrebanList, a 0.0066 points lower volatility. The variable Ban is very significant and shows that the short selling ban has an effect of a higher average price volatility of 0.0277 points than when a firm is not in the ban period. The interaction variable BanList captures the effect of the short selling ban. This variable is highly significant and has a positive coefficient of 0.0138. This means that the short selling ban increases the average price volatility with 0.0138 points. This confirms the hypothesis that the price volatility increases due to the short selling ban is bigger for the banned firms than for the non-banned firms. From figure 2, we can see that there is a peak in price volatility when the short
relationship with the dependent variable price volatility. The first control variable, LnTurnover, has a positive coefficient with a value of 0.0090. This means that when the turnover increases with 1 percent, the price volatility increases with 0.01 x 0.0090 points. So firms with a higher turnover experience higher price volatility. The second control variable is LnMV. The natural logarithm of market value is added in the regression, because it has impact on the price volatility. On the contrary with LnTurnover, LnMV has a negative coefficient of -0.0110. This implies that a 1 percent increase in market value is coupled with a 0.01 x 0.0110 point decrease in price volatility. This means that a higher market value has as a result lower price volatility. 7.3 Globally Systemic Important Banks The third column from Table 2 and the fourth column of Table 3 give the regression analysis where the variables GSIB, PrebanGSIB and BanGSIB are added. These columns are used to examine the effect of a short selling ban on a Globally Systemic Important Bank. First, Table 2 shows the regression with the bid ask spread as the dependent variable. When we take a look at the third column, we see that the variable GSIB is highly significant and has a positive coefficient of 0.2765. The variable List has a negative coefficient of -0.0754. This would suggest that a GSIB firm has a higher bid ask spread and is therefore less liquid and that the other banned firms are more liquid. This seems surprising, since we expected that the GSIB’s are more liquid than the other banned firms. This can be explained by the variable LnTurnover. This variable has a negative coefficient of -0.1322. A Globally Systemic Important Bank has a much higher level of turnover than the other banned firms. Because of the negative coefficient of LnTurnover and the higher level of turnover of the GSIB’s, the bid ask spread for the GSIB’s is lower than for the other banned firms. This means that the GSIB’s have therefore a higher level of liquidity than the other banned firms. The variables Preban and Ban are highly significant with a negative and positive coefficient, respectively. This means that the bid ask spread was lower before the
ban and higher during the ban. The effect of the short selling ban is therefore that the liquidity reduces. When we take a look at the variables BanList and BanGSIB, we see that the variable BanList has a positive coefficient and the variable BanGSIB a negative coefficient. This implies that the effect of the short selling ban is weaker for GSIB’s. This confirms the hypothesis that the effect of a short selling ban is weaker for GSIB’s than for the other banned firms. This is in line with Figure 1 as well, where we can see that at the moment the ban is implemented the bid ask spread increases, but the least for the GSIB’s. Second, table 3 shows the results of regression with the dependent variable volatility. In this regression, different form the regression in table 2, the control variable logarithmic market value is included. All variables used in this regression are highly significant. If we take a look at the fourth column, we can see that both the variables List and GSIB have positive coefficients. The coefficient for the GSIB is higher than for the variable List, which implies that on average the GSIB stock prices are more volatile. The variable Preban has a negative coefficient and the variable Ban a positive one. This means that on average the price volatility was lower before the ban was active. This applies for the variables PrebanGSIB and BanGSIB as well, where the variable PrebanGSIB has a lower coefficient than the variable BanGSIB. This means that the short selling ban had as effect higher average price volatility. When we take a look at the variables BanList and BanGSIB, we see a positive and negative coefficient, respectively. This implies that the ban on short selling has as effect higher price volatility for banned firms than for the GSIB’s. This confirms the hypothesis as well that the effect of the short selling ban is weaker for the GSIB’s.
8. Conclusion
This paper examined the effects of the 2008-2009 short selling ban on the market liquidity and price volatility in the United Kingdom. The stocks listed on the FTSE350 were used to do a regression analysis on. The time period used started at the 1st of January 2008 and ended at the 31st of December 2009 in which the short selling ban was active from the 19th of September 2008 till the 16th of January 2009. In this research 22 banned stocks are compared with the control group, which consists of 321 stocks. Three hypotheses were tested in this paper about the liquidity, measured in the bid ask spread, the volatility and if the effects of the short selling ban were weaker for GSIB’s. Table 3 shows the regression results with dependent variable volatility. These results show that the short selling ban increased the price volatility of banned stocks significantly more than for non-banned stocks. This outcome confirms the first hypothesis that the price volatility increased more for the banned firms, due to the short selling ban, than for the non-banned firms. Figure 2 shows that at the moment the short selling ban is implemented, the price volatility increased more for banned firms. From the regression results from table 2, strong evidence is provided that the short selling ban had as an effect that the bid ask spread, as a measure for liquidity, increased significantly more for the banned firms than for the non-banned firms. Therefore, the second hypothesis is accepted and the banned firms were less liquid than the non-banned firms. This outcome is also in line with Figure 1, where the bid ask spread increased more for the banned firms than for the non-banned firms. The third hypothesis is about the effects of the short selling ban on Globally Systemic Important Banks, GSIB’s. The regression results in the third column of Table 2 and the fourth column of Table 3 provide evidence that the short selling ban had a weaker effect on the GSIB’s than on the other banned firms. The bid ask spread increased the least for the GSIB’s when the short selling ban was implemented. The GSIB’s were, therefore, most liquid during the ban period.
This is in line with Figure 1 where we can see that the bid ask spread is the lowest for the GSIB’s. The price volatility of the GSIB’s increased less, due to the short selling ban, than the other banned firms as well. The GSIB’s and other banned firms experienced mainly a large peak when the short selling ban was implemented and when the ban period ended. Further research in the future could have a look at the effects of a short selling ban implemented in different countries. When the effects are tested in different countries, you could have a look at the stocks price and trading volume as well. Furthermore, there could be looked deeper into the effects of the ban on liquidity and volatility and see if there were external factors that influenced those variables as well. The GSIB’s could be tested in different countries and see if they act in the same way.
References
Beber, A. & Pagano, M. (2013), ‘Short-Selling Bans Around the World: Evidence from the 2007-09 Crisis’, The Journal of Finance, p. 343-381. Bodie, Z., Kane, A. and Marcus, AJ. (2014), ‘Investments’ (10th edition). The McGraw-Hill/Irwin Education. Boehmer, E., C.M. Jones and X. Zhang (2009), ‘Shackling Short Sellers: The 2008 Shorting Ban’. 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. (2005), ‘A study of market-wide short-selling restrictions’ (Working paper 2009- 21)’. Vanderbilt University and Cornell University. Clifton, M., Michayluk, D. (2010), ‘The impact of short selling restrictions and extreme uncertainty on liquidity and order flow: Evidence form the London Stock Exchange.’ Capital Markets CRC, University of Technology, Sydney Diamond, D. W., & Verrecchia, R. E. (1987), ‘Constraints on Short-Selling and Asset Price Adjustment to Private Information’. Journal of Finacial Economics, 18(1987), 277-311. Diether, K., Lee, K-H., Werner, I. (2009), ‘Short-sale strategies and return predictabilities.’ The society for Financial Studies, Oxford University. Financial Services Authority (2009), ‘Short Selling, DP 09/1.’ http://www.fsa.gov.uk/Pages/Library/Policy/DP/2009/09_01.shtml Hansson, F., Fors, E. R. (2009), ‘Get shorty? Market impact of the 2008-09 U.K. short selling ban.’ Working Papers in Economics No 365, University of Gothenburg. Laeven, L., Valencia, F. (2012) ‘Resolution of banking crises: The good, the bad, and the ugly.’ Research Department of the International Monetary Fund and Research Fellow, CEPR. Marsh, I. W., & Payne, R. (2012), ‘Banning short sales and market quality: The UK’s experience.’ Journal of Banking & Finance, 36 (2012) 1975-1986. Miller, E., (1977), ‘Risk, uncertainty, and divergence of opinion’. Journal of Finance, 32, 1151-1168. Moshirian, F., (2012), ‘The future and ynamics of global systemically important banks.’ Journal of Banking & Finance, 36 (2012) 2675-2679.http://www.fsa.gov.uk/pages/Library/Communication/PR/2008/102.shtml
Appendix
Table 2. Bid Ask Spread BidAskSpread (1) (2) (3) List -0.1576*** -0.0198 -0.0754*** (0.0125) (0.0126) (0.0140) Preban -0.0933*** -0.0610*** -0.0603*** (0.0050) (0.0050) (0.0050) PrebanList -0.0003 0.0197 0.0266 (0.0191) (0.0190) (0.0213) Ban 0.1402*** 0.1756*** 0.1761*** (0.0065) (0.0065) (0.0065) BanList 0.1911*** 0.1436*** 0.2110*** (0.0248) (0.0248) (0.0277) LnTurnover -0.1299*** -0.1322*** (0.0012) (0.0012) GSIB 0.2765*** (0.0298) PrebanGSIB -0.0291 (0.0452) BanGSIB -0.3258*** (0.0588) Constant 0.4161*** 1.3090*** 1.3241*** (0.0033) (0.0089) (0.0090) N 180996 174643 174643 R2 0.0094 0.0704 0.0711 Adjusted R2 0.0093 0.0704 0.0711 *, **, *** denotes significance level of 10%, 5% and 1%, respectivelyTable 3. Volatility Volatility (1) (2) (3) (4) List 0.0068*** 0.0032*** 0.0099*** 0.0053*** (0.0004) (0.0004) (0.0004) (0.0005) Preban -0.0013*** -0.0024*** -0.0011*** -0.0010*** (0.0002) (0.0002) (0.0002) (0.0002) PrebanList -0.0041*** -0.0050*** -0.0066*** -0.0045*** (0.0007) (0.0007) (0.0006) (0.0007) Ban 0.0305*** 0.0308*** 0.0277*** 0.0277*** (0.0002) (0.0002) (0.0002) (0.0002) BanList 0.0094*** 0.0111*** 0.0138*** 0.0157*** (0.0009) (0.0009) (0.0008) (0.0009) LnTurnover 0.0037*** 0.0090*** 0.0090*** (0.0000) (0.0001) (0.0001) LnMarketValue -0.0110*** -0.0112*** (0.0001) (0.0001) GSIB 0.0237*** (0.0010) PrebanGSIB -0.0101*** (0.0015) BanGSIB -0.0086*** (0.0019) Constant 0.0390*** 0.0154*** 0.0550*** 0.0573*** (0.0001) (0.0003) (0.0004) (0.0004) N 180996 174643 174643 174643.0000 R2 0.1150 0.1631 0.2634 0.2668 Adjusted R2 0.1150 0.1630 0.2634 0.2668 *, **, *** denotes significance level of 10%, 5% and 1%, respectively
Table 4. List of banned firms Banned firms included in the sample: 1. Aberdeen Asset Management PLC 2. Alliance Trust PLC 3. Aviva PLC 4. Barclays PLC * 5. Brit Insurance Holdings PLC 6. Close Brothers Group PLC 7. F&C Asset Management PLC 8. Friends Provident PLC 9. HSBC Holdings PLC * 10. Investec PLC 11. Legal & General Group PLC 12. Lloyds TSB Group PLC * 13. Old Mutual PLC 14. Provident Financial LTD 15. Prudential PLC 16. Rathbone Brothers PLC 17. Royal Bank of Scotland Group PLC * 18. RSA Insurance Group PLC 19. Schroders PLC 20. St James’s Place PLC 21. Standard Chartered PLC * 22. Standard Life PLC *: Globally Systemic Important Bank Banned firms excluded in the sample: 23. Admiral Group PLC 24. Alliance & Leicester PLC 25. Aviva PLC 26. Bradford & Bingley PLC 27. Chesnara PLC 28. European Islamic Investment Bank PLC 29. HBOS PLC 30. Highway Insurance Group PLC 31. Just Retirement Holdings 32. London Scottish Bank PLC 33. Novae Group PLC 34. Tawa PLC
