University of Amsterdam Faculty of Economics and Business
Department of Finance
Bachelor Thesis
The Impact of the Short Selling Ban in the UK in 2008-2009
Abstract: On the September 18, 2008 in times of market turmoil and financial crisis the Financial Service Authority imposed an emergency short selling ban on 29 financial stocks. According to the FSA it was necessary in order to restore the stability in the market and to protect the institutions from devaluation. This study examines the effect of the short selling ban on both, banned and non-banned stocks. The impact on the financial market is analyzed on the basis of the trading volume, stock price, volatility and bid-ask spreads.
The multivariate regression is performed on these variables. Using the OLS regression we find evidence that in case of the stocks subject to the ban i) the trading volume decreases ii) the stock price return volatility significantly increases suggesting lower performance of banned stocks iii) the stock prices decrease over the three periods and iv) the bid ask spreads narrow. Given the results, FSA does not seem to have achieved the objective to calm the disorderly markets and to protect the quality of the financial sector.
Keywords: short selling, ban, volatility, liquidity, FSA JEL classification: G01, G14, G18
Supervisor Student
Jan Lemmen Adela Scepkova
STATEMENT OF ORIGINALITY
The work contained in this thesis has not been previously submitted for a degree or diploma at any other higher education institution. To the best of my knowledge and belief, the thesis contains no material previously published or written by another person except where due references are made.
Adela Scepkova
1 Introduction
Short selling is frequently considered to be beneficial for the functioning of the market through enhancement of liquidity and more accurate price formation (FSA, 2009, pp.10-11). Most academic literature, apart from being in line with this statement, interprets the short sellers as much better informed than the rest and thus contributing to the efficient pricing (Boehmer, Jones, & Zhang, 2008, p.1365). However, following the collapse of the
investment banks Bear Stearns and Lehman Brothers, the authorities decided to implement restrictions on short selling activity (Bodie, Kane, & Marcus, 2014, p.84). Moreover, given the high and prolonged price volatility and downward pressure on the prices of financial stocks, on September 18 2008 the Financial Service Authority (FSA) in the UK took the step of introducing the temporary short selling measures on an emergency basis (FSA, 2009, p.3). It constrained the creation and increase of net short positions, both, naked and covered, in 29 financial stocks from the London Stock Exchange. In one-month time, an additional six stocks were added and one was removed from the list. On the same day the Securities and Exchange Commission (SEC) in the US instituted a short selling ban on nearly 1000 firms. Altogether, more than 24 countries followed a similar strategy and posted restrictions on short sales varying in intensity, breadth and span (Hansson & Fors, 2009, p.1). The ban in the UK came into force on September 19, 2008 and was effective until January 16, 2009 with a revision after 30 days of the announcement1 (Marsh & Payne, 2012, p.1). The statements from FSA’s senior executives clarify that the ban was introduced due to the extreme market turbulence. According to Hector Sants, the chief executive of the FSA, the idea behind the decision was “to protect the fundamental integrity and quality of markets and to guard against further instability in the financial sector”. Sir Callum McCarthy, the chairman of the FSA, further clarified the FSA’s measures later on the evening of the day of the announcement. A cause for concern was the volatility and “the incoherence of equities” and the danger of financial institutions being subject to the short selling pressures. Additionally, given such movements in prices that could have led to depositors’ uncertainty, they have designed the short selling ban in order to “calm” the market.2
1 Along with other changes, revisions to the list of the banned firms were made. Seven firms were added in a
period of two weeks, and another six firms were delisted during the ban.
2 The full speech can be found at FSA’s website
Despite the fact that the majority of regulators are in favor of the short selling ban, there has been an ongoing debate between the general public, issuers, legislators and politicians with respect to the efficacy of the short sale ban. As a response to this topic, several studies have been conducted, analyzing the effects of the short selling restrictions of 2008. Most of the work has been based on Miller (1977). His findings suggest that the short selling activity puts upward pressure on the stock prices. More specifically, it leads to overpricing because the securities reflect more information of positive investors while the pessimistic, more rational investors are being excluded from the market.
The purpose of this research paper is to empirically assess, on the basis of the stock price volatility and trading volume changes, to what extent the FSA was successful in order to restore investors’ confidence in public capital markets and to protect financial institutions from rapid devaluation. In this paper the stocks from FTSE 100 are examined. The sample consists of data of 93 companies, the rest of the companies that were not a part of the FTSE 100 index, were excluded. The data covers a period from May 28, 2008 until May 13, 2009. For each stock we indicate whether it was subject to the ban in this period or not. The banned stocks will be compared to the rest of the companies not affected by the ban. The main focus is on the effects on trading volume, volatility, stock price and bid-ask spreads, but market value and stock price return of each company will also be taken into consideration.
In line with the previous studies, the evidence shows that the trading activity decreases with the imposition of the short selling ban. Furthermore, the volatility increases significantly for the banned sample compared to the non-financials implying inferior performance. The results further indicate that the stock prices are lower for the banned firms and that bid-ask spreads narrow over the period the ban was imposed, suggesting higher liquidity.
The remainder of this paper is organized as follows: In section 2 the process of short selling will be explained, followed by a review of relevant literature and development of the hypothesis of this paper. Section 3 provides details of the methodology needed to test the hypotheses. The data used can be found in Section 4. Section 5 describes the results obtained from a multivariate regression. The conclusion will be formulated in the final section.
2 Theoretical framework
2.1 Time Line of Events
The financial crisis first started in the United States but quickly affected many other countries in the world. The financial turmoil has led to the introduction of emergency short selling measures. The first authority that took a step towards this type of restrictions was the SEC in the US. In July 2008, the SEC introduced a ban affecting 19 financial stocks. After the expiration in mid-August, the SEC has decided to implement another ban effective in all the US stocks (Boehmer et al, 2008, pp. 1363-1368). On the 18th of September 2008 the shorting ban in the UK was implemented. The FSA has named 28 financial companies that would fall under the ban. Consequently, on the 22nd September another seven companies fell under the short selling ban restrictions and five other stocks were excluded from the ban list later on. The complete list of banned firms can be found in Table 4 in the appendix. Furthermore, the ban was accompanied by disclosure requirements that were supposed to prevent potential market abuse. Both, the ban itself and the disclosure requirements were scheduled to be effective until 16 January 2009 (FSA, 2009, pp.3, 29).
2.2 Background
Short selling, by definition, is a sale of a financial instrument that is not owned by the investor at the time of the sale. Typically, the seller borrows the security and later
repurchases it in order to return it to its original owner. This is the so-called ‘covered short selling’. The investor is motivated by the belief that the price of the financial instrument will decline in the future, hence realizing a profit from the difference between the original price and the price of the security at the settlement date (Culp & Heaton, 2008, p.46). In case of the ‘naked short sale’ no security has been borrowed. This can lead to the so - called ‘failure to deliver’, when the short seller does not purchase the stock before the delivery date. Therefore, given such a risk, many countries have prohibited it or have posed restrictions on this type of short selling (Helmes, Henker, & Henker, 2010, p.5).
Covered short sale involves a number of transactions. Initially, the investor borrows the shares that he is planning to short from a long-term shareholder, for example an insurance firm or a pension fund, for a fee. In the second stage, the seller short sells the shares. Lastly, the investor has to repurchase identical securities so he can return them to their true owner (FSA, 2009, p.6).
At times of normal market conditions, short selling is considered to improve the efficient functioning of the market through the increase of trading volume and reduction of transaction costs. Moreover, by increasing the number of sellers in the market it contributes to the
liquidity (FSA, 2009, p.10). This fact is measured by the bid-ask spread. The narrower the spread, the more liquidity is expected for the particular stock. In contrast, market regulators, especially in times of market turmoil, often impose restrictions on short selling activity. One of the factors is the potential negative effect on volatility. More specifically, in case of a bad news announcement short sellers increase their positions. This increase in the volume of trading could put excessive downward pressure on share prices (Mckenzie & Henry, 2007, pp.1-2). Another argument against short selling is that it leads to transparency deficiencies. This problem arises from the difference of the amount of information between the short sellers, which are informed, and the uninformed market participants. The short sellers are in general, not interested in sharing their private information. Moreover, these information asymmetries could be a cause of price inefficiency. For this reason many financial institutions have imposed the disclosure requirements. However, high transparency may lead to lower liquidity because traders do not want to reveal their strategies (FSA, 2009, pp.11-14). 2.3 Literature Review
Short sales, in general, play an important role in the financial market. There have been a number of studies concentrating on different aspects of the short selling restrictions. For example, Diamond and Verrecchia (1987), based on their theoretical model predict, a
decrease in liquidity under the shorting ban as prices are adjusting slower to new information. While other papers, in majority, concentrate on individual stocks and on how do the short selling regulations affect them, Charoenrook and Daouk (2005) analyze 111 countries under short selling restrictions and perform analysis of aggregate market returns. They find that the non-short selling markets are more liquid and the index returns less volatile. They conclude that the ability to sell short is profitable and contributes to the market quality. In contrast, Diether, Werner, and Lee (2009) argue that the ability to short sell does not contribute to market quality. Furthermore, they provide evidence that the intraday volatility increases when the short selling restrictions are lifted and the effect on liquidity of the market is insignificant. Clifton and Snape (2008) perform analysis of the behavior of a sample of banned and non-banned, the so-called, control stocks in the UK. They find that the average bid-ask spreads rise for all, nevertheless it affects the banned stocks by 150% more than the control sample. They reach a conclusion that the ban led to a decline in trades, volume and turnover, and that
the stocks restricted from short selling experience a deterioration in liquidity. In line with this conclusion Hansson and Fors (2009) study the market impact of the 2008-09 UK short selling ban in terms of returns and market quality. They do not find any evidence of calmer market conditions. However, what they do find is a significant deterioration in market quality. Findings of most of the papers mentioned above suggest that the short selling restrictions have rather a detrimental effect. In apparent contrast from these studies, Shkilko, Van Ness and Van Ness (2009) provide evidence that short selling may cause occasional excessive price pressure. They point out that the short sellers are often aggressive during the days of large negative price reversals, which severely affects the magnitude of price declines and the amount of liquidity. Allen and Gale (1991) claim that short selling has a destabilizing effect on the economy and when short selling activity is not allowed the firms are competitive and the market is efficient.
Closest to the analysis of this paper is the work of Boehmer, Jones and Zhang (2008). They study the effects of the short selling ban imposed by the SEC in the US on September 2008.
They analyze what consequences does the US ban have on the overall market quality, short selling activity, the stock prices and the aggressiveness of the short sellers on the basis of intraday data of bid-ask spreads, volatility and trading volume. In their paper they perform the firm-pair fixed effects panel regression. They regress the market capitalization, dollar trading volume, the proportional daily range of transaction prices, and the daily volume-weighted average share price on the difference in quantities of the short sales, and various types of spreads, such as time-weighted and trade-weighted relative effective, between
banned and control samples. They find that all but small capitalization firms are subject to the severe market quality degradation. Furthermore, they provide evidence that the firms with short selling restrictions beforehand were less affected. These firms are the so-called hard-to-borrow stocks whose availability is often limited and which earn a lower, sometimes even negative, credit rate on short sale proceeds. Moreover, they show that the short sellers are in majority much better informed than the rest of the traders and thus contribute to the efficient pricing. Their data, concentrating on nearly 1000 US banned stocks, is much broader
compared to the data used in this paper which focuses on 14 banned stocks. This research will use similar analysis, however applied to a different, more homogeneous, market. In contrast to this paper, Boehmer et al. (2008) perform the panel regression and analyze different variables.
2.4 Hypotheses
Based on the summary of the literature, the following hypotheses are addressed. Some of the trading activity not being longer allowed during the short selling ban is expected to cause a decline in overall trade. Hence, the trading volume is expected to be negatively affected if stocks are banned from short selling. The first hypothesis is therefore:
Hypothesis 1: The ban on short selling decreases trading activity of short sale constrained stocks.
Chang, Cheng and Yu (2007) find that no restrictions to short selling activity give rise to higher daily returns volatility. Notwithstanding, the evidence of the short selling bans’ effects on volatility is inconclusive. In contrast to the above-mentioned study, Conrad (1989),
Charoenrook and Daouk (2005) discover that the imposition of short selling ban on stocks leads to a rise in volatility. In line with this suggestion, the empirical study of Scheinkman and Xiong (2003) concludes that the difficulty of being able to short sell leads to the rise in bubbles and significantly higher price volatility. Therefore, the second hypothesis is formed as follows:
Hypothesis 2: The volatility for the banned stocks will be higher compared to the control sample.
Research done by Miller (1977) suggests that in absence of short selling the price of a security increases if opinions of different investors are heterogeneous. He finds that short selling measures lead to the so - called overpricing effect. This claim is explained by the fact that if short selling is prohibited, the stock prices consist of only the valuations of bullish investors and bearish investors that own the stock. The rest of bearish investors that do not own any stock do not affect the prices because they are excluded from the trading. Therefore, this results in a price increase above their full information values under the ban. Harris, Namvar and Phillips (2009) find that short selling ban in the US in 2008 imposed by the SEC inflated the price by 10-12% of the banned stocks relative to the non-banned sample. They claim that given the price increase, buyers were paying more to obtain the banned stocks during the ban, which led to the $2.6 billion wealth transfer from buyers to sellers. Such a substantial amount could be a threat to maintain fair markets. Thus, the third hypothesis is stated as:
Diamond and Verrecchia (1987) test on their model the effects of the ban on the speed of adjustments of security prices and also the effect on liquidity through the bid-ask spreads. They find that short selling restrictions lead to a decline in the speed of adjustment of prices to private information. Examining the bid-ask spread, they provide evidence that if both the informed and uninformed investors are subject to the ban, the bid-ask spread stays the same. In contrast, Helmes et al. (2010) and Beber and Pagano (2013) suggest that banned stocks experience an increase in bid-ask spreads, which indicates a significant decline in market liquidity. The fourth hypothesis is therefore stated as follows:
3 Methodology
3.1 Trading volume
To measure the first of the above-mentioned hypotheses we perform a regression with turnover as the dependent variable. Turnover is calculated as a number of shares traded on a daily basis for each stock i. Moreover, to examine whether the trading activity was severely affected due to the restrictions, the following regression is performed:
𝑌𝑡 = 𝛼0+ 𝑎1𝐷𝑖,𝑡𝐶+𝑎2𝐷𝑖,𝑡𝐹 × 𝐷𝑖,𝑡𝑃𝑟𝑒+ 𝑎3𝐷𝑖,𝑡𝐹 × 𝐷𝑖,𝑡𝐵𝑎𝑛+ 𝑎4𝑃 + 𝑐𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 + 𝜀𝑖,𝑡 (1)
On the left side there is the dependent variable 𝑌𝑡, which represents the turnover. On the other side 𝛼0 is a constant; the dummy variable 𝐷𝑖,𝑡𝐶 equals to 1 if the observation is part of the nonfinancial sample, which is not subjected to the ban. Another dummy variable 𝐷𝑖,𝑡𝐹 takes the value of 1 if the observation is a financial firm and therefore on a to-be-banned list. Similarly, two cross terms are used: 𝐷𝑖,𝑡𝑃𝑟𝑒 and 𝐷𝑖,𝑡𝐵𝑎𝑛,which are equal to 1 for the pre-ban period and ban period, respectively. These dummy variables are 0 otherwise. If 𝑎2 and 𝑎3 are significantly different from 0 it reveals that the behaviour of banned and control sample stocks during a particular period differs. Variable P stands for the stock price for a particular firm. Also included is the control variable, which stands for the market value of each firm constructed as the share price multiplied by the number of ordinary shares in issue. Given that market value and turnover are right-skewed and always greater than zero, in the regression we take the natural logarithm of these variables.
3.2 Volatility
In order to examine the second hypothesis we will look at the effects on the stock return volatility. Following Boehmer et al. (2008), the return volatility is used to analyze whether the imposed ban was successful.
The analysis is based on a multivariate regression equation:
𝑌𝑡= 𝛼0+∝1 𝐷𝑖,𝑡𝐶 + 𝑎2𝐷𝑖,𝑡𝐹 × 𝐷𝑖,𝑡𝑃𝑟𝑒+ 𝑎3𝐷𝑖,𝑡𝐹 × 𝐷𝑖,𝑡𝐵𝑎𝑛+ 𝑐𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 + 𝜀𝑖,𝑡 (2) The dependent variable is the monthly volatility of returns of a stock i at time t, which serves as a proxy for the risk. On the right hand side, the constant and dummy variables are the same as in equation (1) above. As the control variables the market value, turnover, stock price and return are used. The return on a stock i is measured in basis points.
3.3 Stock Price
The analysis of the third hypothesis will be performed similarly as in case of the first two hypotheses above. We will examine the effect of the short selling restrictions on the stock price through the regression (3).
𝑌𝑡= 𝛼0 +∝1𝐷𝑖,𝑡𝐹 + 𝑎2𝐷𝑖,𝑡𝐹 × 𝐷𝑖,𝑡𝑃𝑟𝑒+ 𝑎3𝐷𝑖,𝑡𝐹 × 𝐷𝑖,𝑡𝐵𝑎𝑛+ 𝑐𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 + 𝜀𝑖,𝑡 (3)
The dependent variable in this case is the stock price, which is regressed on market value, turnover, volatility and return. The rest of the equation remains unchanged.
3.4 Bid-Ask Spread
Finally, we test the hypothesis 4. The effect on bid-ask spread is examined. The bid-ask spread will proxy the effects of the ban on liquidity and it is computed as follows:
Percentage Bid-Ask Spread= 100* (Ask price-Bid price
Ask price )
A simple regression will be performed:
𝑌𝑡= 𝛼0+ 𝛼1𝐷𝑖,𝑡𝐶 + 𝑎2𝐷𝑖,𝑡𝐹 × 𝐷𝑖,𝑡𝑃𝑟𝑒+ 𝑎3𝐷𝑖,𝑡𝐹 × 𝐷𝑖,𝑡𝐵𝑎𝑛+ 𝑎4log(𝑡𝑢𝑟𝑛) + 𝜀𝑖,𝑡 (4)
Where 𝑌𝑡 represents the dependent variable, in this case the bid-ask spread. 𝛼0 is the constant and the dummy variables used are same as in equation (1) . The independent variable is the logarithmic turnover and the last term, 𝜀𝑖,𝑡, represents the random error term.
For each regression the heteroscedasticity is tested in order to get more reliable results. The term homoscedasticity refers to the sample in which the variances across observations differ and therefore the following holds:
𝑉𝑎𝑟(𝑢𝑡) = 𝜎𝑡2
This equation implies that each observation from the sample has a different variance possibly leading to biased results. For each regression the Breusch-Pagan test was performed in order to find whether the heteroscedasticity is present. High value of chi squared obtained from the tests confirms the presence of the heteroscedasticity. Given this, the robust standard errors are used in the regressions relaxing the assumptions of the errors to be independent and
4 Data
The analysis of this paper covers the period from 28 May 2008 and extends to 13 May 2009. The time span is divided into three equal periods of 82 trading days:
a) the period from 28 May 2008 until 18 September 2008 called the pre-ban period; b) the period from 19 September 2008 to 15 January 2009 called the ban period; and c) the period from 16 January 2009 to 13 May 2009 called the post-ban period.
To be included in the sample, stocks have to be a part of FTSE 100 for the whole period from 28 May 2008 until 13 May 2009, covering the pre-ban, ban and post-ban period. The daily data is extracted from the Datastream database for FTSE100 index including stock prices, turnover, bid and ask prices and market value of each company. The returns are calculated manually using the stock prices. The dates and more specifications of the short selling restrictions are retrieved from the FSA’s website (Appendix, Table 4). This information allows distinguishing between banned stocks and the control sample. After applying the criteria to the filter, the sample consists of 93 stocks. The banned group includes 14 FTSE100 stocks that are short selling restricted. Consequently, 79 companies not affected by the ban, are classified as a control sample. Therefore, as of 28 May 2008 nearly 17% of the stocks of the index were under short selling measures. During the analysis the banned sample is compared to the stocks that do not fall under the ban, to see the differences and the effect of the short selling ban on the market.
The analysis is based on the assumption that the short selling decreases with the imposition of the ban significantly. The dataset does not consist of any kind of information about short selling activity, for example the percentage of shorting of the total trading volume or the short selling interest etc. Therefore, it is not clear what level of short sales is present during a particular day or period.
Table 1 provides summary statistics for both, the control sample as well as the banned stocks, for three different periods – pre-ban, ban and post-ban. The sample consists of 14 banned and 79 control stocks. All the variables used in the regression are summarized. Also, because market value and turnover are skewed, the natural logarithms of these variables are used. From the table we can see that the trading activity, measured by the turnover, decreases during the ban period for the stocks subject to the short selling measures whereas it stays
roughly the same for the control sample. After the ban has been lifted the trading of the banned firms rose again. The volatility was also affected by the short selling measures. In the ban period it increased for the financials by 0.02 points on average, however it decreased for the rest of the stocks. The table also shows that the price for the financial stocks has
experienced a decline over the sample time frame, decreasing even further in the post-ban period. By looking at the bid-ask variable it can be concluded that the liquidity in case of banned stocks was reduced during the ban period but the control sample did not experience any significant change.
Table 1. Descriptive statistics Financials
a) pre-ban period
Obs Mean Median Std.Dev. Min Max Skewness
logMktV 1134 8.99 8.65 1.14 7.51 11.60 0.57 logturn 1134 9.15 9.36 1.50 5.56 13.47 -0.32 P 1134 619.55 535.00 553.81 69.30 2526.83 1.45 ret 1134 0.00 0.00 0.05 -1.00 0.13 -9.37 Vol 1134 0.04 0.03 0.03 0.01 0.58 12.61 Bid-ask 1134 0.001 0.00 0.00 0.00 0.02 3.89 b) ban period
Obs Mean Median Std.Dev. Min Max Skewness
logMktV 1148 8.67 8.46 1.08 7.07 11.62 0.93 logturn 1148 15.58 15.80 1.60 9.53 20.11 -0.42 P 1148 445.37 379.00 349.30 39.17 2205.00 1.05 ret 1148 1.05 0.00 0.07 -0.39 0.32 0.44 Vol 1148 0.06 0.06 0.03 0.02 0.19 1.52 Bid-ask 1148 0.005 0.00 0.02 0.00 0.50 24.48 c) post-ban period
Obs Mean Median Std.Dev. Min Max Skewness
logMktV 1134 8.18 7.86 0.93 6.91 10.66 1.25 logturn 1134 16.03 16.28 1.67 12.25 19.95 -0.20 P 1134 361.25 272.63 295.35 19.99 1240.87 0.68 ret 1134 0.00 0.00 0.07 -0.67 0.50 -0.36 Vol 1134 0.06 0.05 0.03 -0.02 0.19 1.22 Bid-ask 1134 0.002 0.00 0.01 -0.23 0.17 -29.34
Nonfinancials
a) pre-ban period
Obs Mean Median Std.Dev. Min Max Skewness
logMktV 21557 8.64 8.39 1.12 5.77 11.66 0.63 logturn 21557 8.68 8.64 1.22 0.83 13.73 -0.04 P 21557 713.27 528.21 617.56 19.99 5105.32 1.93 ret 21557 0.00 0.00 0.04 -1.00 0.73 -0.20 Vol 21556 0.04 0.03 0.02 -0.02 0.22 2.34 Bid-ask 21555 0.00 0.00 0.00 -0.23 0.50 51.22 b) ban period
Obs Mean Median Std.Dev. Min Max Skewness
logMktV 21543 8.66 8.41 1.12 5.77 11.66 0.61 logturn 21543 15.61 15.56 1.22 7.74 20.63 0.01 P 21543 722.61 535.00 622.75 19.99 5105.32 1.88 ret 21543 0.00 0.00 0.04 -1.00 0.73 0.91 Vol 21543 0.03 0.03 0.02 -0.02 0.58 4.15 Bid-ask 21541 0.00 0.00 0.00 -0.23 0.14 -7.18 c) post-ban period
Obs Mean Median Std.Dev. Min Max Skewness
logMktV 21557 8.68 8.44 1.12 5.77 11.66 0.60 logturn 21557 15.59 15.55 1.21 7.74 20.63 -.0.65 P 21557 726.86 540.50 621.83 39.17 5105.32 1.88 ret 21557 0.00 0.00 0.04 -1.00 0.73 -0.79 Vol 21556 0.04 0.03 0.02 0.00 0.58 4.05 Bid-ask 21555 0.00 0.00 0.00 0.00 0.50 67.94
5 Results
The Table 2 below reports results from the pooled OLS regression on trading volume (Panel A), volatility (Panel B), stock price (Panel C) and bid-ask spread (Panel D).
The coefficient estimates are given on the left hand side in each panel and p-values on the right.
Table 2.
OLS Regression
The following tables report results for the pooled OLS regression analysis on 93 FTSE100 stocks for the period from 28 May 2008 to 13 May 2009.
Panel A
Turnover as dependent variable
Coef p-Values C -0.384 0.000 FPRE -0.160 0.007 FBAN -0.617 0.000 P -0.001 0.000 logMtkV 0.508 0.000 _cons 5.27 0.000 R-squared 0.2704 n of observations 22691
Log(turn) represents the logarithmic turnover and the dependent variable in this case. The dummy variable C is equal to 1 if the firm is not subject to the ban and 0 otherwise. The other dummy variables FPRE and FBAN are equal to 1 if the firm was banned in the pre-ban or ban period, respectively. LogMktV stands for the logarithmic market value of each stock i.
Panel B Panel C Vol as dependent var. P as dependent var.
Coef p-Values Coef p-Values
C -0.0182 0.000 172.4896 0.000 FPRE -0.0177 0.000 65.1458 0.000 FBAN 0.0112 0.000 -120.5326 0.004 logMktV -0.0031 0.000 243.4235 0.000 logturn 0.0042 0.000 - 202.7694 0.000 P -3.83E-07 0.107 Vol - 225.6333 0.105 ret -0.0239 0.085 64.9527 0.353 _cons 0.0435 0.000 203.6080 0.000 R-squared 0.1633 0.2740 N of observations 22690 22690
Vol is the dependent variable of Panel B and represents the stock return volatility. In Panel C, the stock price P serves as the dependent variable. C is equal to 1 if the firm is not subject to the ban and 0 otherwise. The two cross term variables FPRE and FBAN are equal if the firm was banned in the pre-ban or ban period, respectively. LogMktV stands for the logarithmic market value of each stock i. Logturn is the logarithm of turnover representing the trading volume and ret stands for the stock price return.
Panel D
Bid-Ask spread as dependent variable
Coef p-Values C -0.001 0.000 FPRE -0.001 0.000 FBAN 0.002 0.000 logturn 0.000 0.000 _cons 0.006 0.000 R-squared 0.0359 n of observations 22689
Bid-ask spread is used as the dependent variable and it is regressed on the following variables. C is equal to 1 if the firm is not subject to the ban and 0 otherwise. The two cross
term variables FPRE and FBAN are equal if the firm was banned in the pre-ban or ban period, respectively. Logturn is the logarithm of turnover representing the trading volume and ret stands for the stock price return.
In Panel A the logarithmic turnover represents the dependent variable. From the results it can be concluded that the trading activity for banned firms decreases prior to the introduction of the short selling restrictions by 0.160 and decreased further by 0.617 points during the ban period. By examining the coefficient of the dummy variable C, we can conclude that the turnover for the control sample decreases throughout the whole sample period. Furthermore, the coefficient of the P variable indicates that for each one-point increase in P, the turnover
decreases by 0.001 points. Also, the coefficient of variable logMktV indicates a positive relation with turnover. Figure 1 elaborates further on these effects.
Furthermore, the descriptive statistics for the turnover are summarized in Table 3. The results provide evidence of the changes in investor activity for all three periods. The table shows that the overall trading activity declined by 18.82% with the introduction of the short selling measures. This decline could be explained by the fact that some of the trading activity was prohibited during the ban or much more expensive and therefore less attractive.
Table 3.
Turnover change
From the table below we can see the fluctuations in trading activity. The table is divided into three parts, representing financials, nonfinancial sample as well as the total. The left side of the table consists of values of turnover over the three periods and their percentage change for the banned firms. It can be seen that the turnover changed severely. With the introduction of the short selling measures the trading activity decreased by 31.09% and started to increase right after the ban was lifted, exceeding the value of the pre-ban period. In contrast, the middle, representing the control sample, shows different results. The results suggest that the change in trading activity in this case is rather small. By looking at the right hand side, we can see that the overall activity decreased by 18.82% and increased by 37.20% when the ban was imposed and lifted, respectively.
Mean % change Mean % change Total % change
PRE, F 23,895.64 PRE, nF 13,501.83 37,397.47
BAN, F 16,467.13 -31.09% BAN, nF 13,890.93 2.88% 30,358.06 -18.82%
POST, F 28,384.72 72.37% POST, nF 13,265.68 -4.50% 41,650.40 37.20%
PRE, BAN, POST stands for the pre-ban, ban and post-ban periods, respectively. F represents the financials and nF the nonfinancial sample.
Figure 1.
Trading volume
The figure shows the effect of the ban on the banned and the control sample as well as the average. Once the ban was implemented the trading volume experienced a decrease in trading volume in both samples.
The dependent variable for Panel B is the monthly return volatility that serves as a proxy for risk. The return volatility decreases by 0.0182 for the nonfinancials, holding other variables constant. The banned sample, on the other hand, reveals a different behavior. Its volatility decreases prior to the ban, however, increases by 0.0112 during the ban period. These results are supported by the graph in Figure 2. All but the stock price variable display significant results given the p-values lower than one percent. The insignificant results between the price and the volatility could be influenced by the fact that during the sample period there were other drivers for price such as the emergency short selling measures affecting the market, possibly also dividends, inflation and other factors.
The effect of the restrictions on the volatility can be observed also from the descriptive statistics table in the Section IV. The stock return volatility of 3.69% in the pre-ban period rises to significantly higher levels during the short selling ban of 6.48% and starts to decline only once the ban is lifted. With regard to the differences between the two groups, the banned sample exhibits lower performance given by the higher average volatility in the ban-period
0 10000 20000 30000 40000 50000 60000 70000 80000 ban control average Ban introduced 19/09/08 Ban lifted on 18/01/09
compared to the 3.13% of the control sample. One of the explanations could be that the period is consistent with high information revelation.
Figure 2. Volatility
This graph shows the effect of the ban on monthly stock return volatility. The graph is divided into three parts to examine each period separately. Furthermore, the graph shows not only the behavior of the banned sample and the control sample but also the average.
In Panel C stock price is the dependent variable. In this case control firms experience an increase in stock price by 172.5 over the whole sample. From the regression results we can further see that the stock price has a negative relation with the FBAN dummy variable, which represents the sample of banned stocks during the ban period. The restricted firms at first experience an increase in the stock price but when the ban is effective it decreases. Therefore, stock prices of the financial firms have experienced a decline over the ban period and started to rise again once the ban was lifted. The changes in price are further explained by the graph in Figure 3. The results of Panel C do not go in line with the findings of Miller (1977) who claimed that in times of short selling measures there is a presence of the overpricing effect. Majority of the variables are statistically significant with p-value lower than one percent and two variables show to have statistically insignificant results. These insignificant results
0 0,01 0,02 0,03 0,04 0,05 0,06 0,07 0,08 0,09 0,1 banned control average Ban introduced on 19/09/08 Ban lifted on 18/01/09
indicate that the independent variables have little or no effect on the dependent-the stock price. That means that during the sample periods the price fluctuations were most probably driven by other factors rather than volatility and return.
Figure 3.
Stock price movement of the three periods
This figure shows the average daily stock prices for banned and control groups for each day from 28 January 2008 until 13 May 2009.
Panel D shows how the ban influenced the market liquidity on the basis of bid-ask spreads. The nonfinancials experience an overall slight decrease in the spread unlike the financial stocks that encounter an increase once the ban is implemented leading to a lower liquidity. This goes in line with the findings of Diamond and Verrecchia (1987), Helmes et al. (2010) and Beber and Pagano (2013). The impact on the bid-ask spreads can be also seen in Figure 4 below. 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 1050 2 8 -5 -2 0 0 8 1 1 -6 -2 0 0 8 2 5 -6 -2 0 0 8 9 -7 -2 0 0 8 2 3 -7 -2 0 0 8 6 -8 -2 0 0 8 2 0 -8 -2 0 0 8 3 -9 -2 0 0 8 1 7 -9 -2 0 0 8 1 -1 0 -2 0 0 8 1 5 -1 0 -2 0 0 8 2 9 -1 0 -2 0 0 8 1 2 -1 1 -2 0 0 8 2 6 -1 1 -2 0 0 8 1 0 -1 2 -2 0 0 8 2 4 -1 2 -2 0 0 8 7 -1 -2 0 0 9 2 1 -1 -2 0 0 9 4 -2 -2 0 0 9 1 8 -2 -2 0 0 9 4 -3 -2 0 0 9 1 8 -3 -2 0 0 9 1 -4 -2 0 0 9 1 5 -4 -2 0 0 9 2 9 -4 -2 0 0 9 1 3 -5 -2 0 0 9 P banned P control Ban introduced on 19/09/08 18/01/09 ban lifted
Figure 4.
Bid – Ask spreads
This figure shows the bid-ask spreads over the 13-month period. The graph is divided into three parts, representing the pre-ban, ban and post-ban periods.
The test for multicollinearity for each regression was also performed. It can be assumed that given the sufficient sample size of 22,690 observations and relatively low level of
correlations between the variables, the multicollinearity is not present. 0 0,02 0,04 0,06 0,08 0,1 0,12 0,14 0,16 0,18 2 8 -5 -2 0 0 8 2 8 -6 -2 0 0 8 2 8 -7 -2 0 0 8 2 8 -8 -2 0 0 8 28 -9 -2 00 8 2 8 -1 0 -2 0 0 8 2 8 -1 1 -2 0 0 8 2 8 -1 2 -2 0 0 8 2 8 -1 -2 0 0 9 28 -2 -2 00 9 3 1 -3 -2 0 0 9 3 0 -4 -2 0 0 9 ban control average Ban introduced on 19/09/08 Ban lifted on 18/01/09
6 Conclusion
This study conducts an analysis of the effect of the short selling emergency ban on the UK market. On September 18, 2008 FSA introduced a short selling ban on 29 stocks, which lasted for 82 trading days. The focus of this paper is concentrated mainly on the consequence of the ban on the market in general, and to see if the financial institutions experienced rapid devaluation. Four hypotheses were formed: (1) The ban on short selling decreases trading activity of short-sale constrained stocks, (2) the volatility for the banned stocks will be higher compare to the control sample, (3) the price of banned stocks is higher than that of non-banned stocks and (4) the bid-ask spread of non-banned sample will widen.
To test these hypotheses the multivariate regressions on the stocks of FTSE100 were
performed and the results of banned stocks were compared to the control sample in different periods. Various measures were analyzed such as turnover, volatility of returns, stock price and bid - ask spreads for the banned and the control group and compared over the three- period time span.
The conclusions are based on the sample of 93 firms over the period of 246 trading days including an 82-day ban period. The time span is represented by both the market turmoil with downward pressure on stock prices and high and prolonged price volatility as well as a more normal market condition by the end of the sample period.
The regression (1) provides strong evidence that the stocks subject to the ban have
experienced significant decrease in the trading activity as the ban has prevented traders from actively participating in the market. Therefore hypothesis 1 is true. Also, the large increase in stock return volatility of the banned sample implies drop in the performance relative to the nonfinancial firms. This indicates that the second hypothesis is accepted as well. The results of the regression analysis in Panel C and the Figure 3 provide evidence for rejecting the third hypothesis. The banned stocks in fact experience higher prices than the control sample. Moreover, the results obtained through the regression (3) suggest that FSA has failed in achieving their goal of preventing the sharp decline of prices. Lastly, the effects on the liquidity were examined through the bid-ask spread. Given the obtained results, the fourth hypothesis is accepted. This yields a decrease in liquidity.
Further research could make use of different measures such as abnormal returns, transaction costs and short selling volume and make a distinction between large and small capital firms.
This would give more insight on the influences of these variables on the financial market resulting from the shorting restrictions. Also, the UK firms could be tested against not only firms from the same index but also across the countries and therefore, deepening the knowledge of the ban effects on financial markets. Furthermore, future papers could
concentrate on the periods in more detail in order to see whether the effect might have been influenced by external factors. The behavior of the market participants to different changes in the financial sector could be taken into consideration analyzing whether their responses were rather slow or rapid.
Since a smaller sample the FTSE100 was used, some of the stocks banned by the FSA were not included in the analysis. This could have resulted in biases in the results. Also, as used in other studies, intraday format seems as a reliable form of data that takes into account price movements throughout the day that could be important in the further analysis.
It can be assumed that some of the confidence was restored by only the fact that the
government and the FSA took some action in times of financial turmoil. Moreover, there is no evidence of what could have happened if the temporary short selling ban was not
implemented at all.
However, given our analysis and results it can be concluded that even though the FSA was aiming to ‘calm’ the disorderly markets and to protect quality of the financial sector and to prevent against its further instability, the short selling ban did not appear to improve the market conditions. To conclude, based on the results from the analysis, the FSA did not seem to achieve their goals of restoring investors’ confidence and to protect the financial institution from rapid devaluation.
References
Allen, F., & Gale, D. (1991). Arbitrage, Short Sales, and Financial Innovation. Econometrica. 59(4), 1041-1068.
Beber, A., & Pagano, M. (2013). Short-Selling Bans Around the World: Evidence from the 2007-09 Crisis. The Journal of Finance. 68(1), 343-381
Bodie, Z., Kane, A., Marcus, A.J. (2014). Investments (10th ed.). McGraw Hill 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. 2005, The world price of Short Selling (Working paper 2009- 21). Vanderbilt University and Cornell University.
Clifton, M., & Snape, M. (2008). The Effect of Short-selling Restrictions on Liquidity: Evidence from the London Stock Exchange (Working paper). Capital Markets Cooperative Research Centre.
Conrad, J. (1989). The Price Effect of Option Introduction. The Journal of Finance, 44(2), 487-498.
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. B., Lee, K., & Werner, I. (2009), Short-sale Strategies and Return Predictability, Review of Financial Studies, 22(2), 575-607.
Financial Service Authority (FSA), (2009). Short Selling. The Financial Service
Authority. DP 09/1, 1-74. Retrieved from the FSA’s website: http://www.fsa.gov.uk. Hansson, F., and Fors, E. R. (2009). Get Shorty? Impact of the 2008 - 2009 U.K. Short
Selling Ban. Working papers in Economics University of Gothenburg 365.
Harris, L., Namvar, E., & Phillips, B. (2009). Price Inflation and Wealth Transfer During the 2008 SEC Short-Sale Ban. The Journal of Investment Management. ). Retrieved from the Social Science Research Network website: http://ssrn.com/abstract=1364390. Helmes, U., Henker, J., Henker, T., 2009. The effect of the ban on short selling on
market efficiency and volatility (Working paper). Retrieved from the Social Science Research Network website: http://ssrn.com/abstract=1568435.
experience. Journal of Banking & Finance, 36(2012), 1975-1986.
McKenzie, M., & Henry, Ó.T. (2007).The Determinants of Short Selling: Evidence from the Hong Kong Equity Market, Accounting and Finance, 52(1), 183-216. Miller, E., (1977). Risk, uncertainty, and divergence of opinion. Journal of Finance,
32, 1151-1168.
Scheinkman, J.A., & Xiong, W. (2003). Overconfidence and Speculative Bubbles. Journal of Political Economy, 111(6), 1183-1219.
Shkilko, A.V., Van Ness, B.F., Van Ness, R.A., 2009. Aggressive short selling and price reversals (Working paper). Retrieved from the Social Science Research Network website: http://ssrn.com/abstract=971210.
Appendix
List of companies subject to the short selling ban Table 4.
FSA amended list, as at 30 September 2008, of UK incorporated banks and insurers.
In total 93 financial companies from FTSE100 - 14 stocks subject to the ban, 79 stocks part of the control sample not affected by the ban
(* if banned)
1. ADMIRAL GROUP* 48. LLOYDS BANKING GROUP*
2. ALLIANCE TRUST* 49. LONDON STOCK EX.GROUP 3. AMEC FOSTER WHEELER 50. MAN GROUP
4. ANGLO AMERICAN 51. MARKS & SPENCER GROUP
5. ANTOFAGASTA 52. MORRISON(WM)SPMKTS.
6. ASSOCIATED BRIT.FOODS 53. NATIONAL GRID
7. ASTRAZENECA 54. NEXT
8. AVIVA* 55. OLD MUTUAL*
9. BAE SYSTEMS 56. PEARSON
10. BARCLAYS 57. PRUDENTIAL*
11. BG GROUP 58. RECKITT BENCKISER GROUP
12. BHP BILLITON 59. REED ELSEVIER
13. BP 60. REXAM
14. BRITISH AIRWAYS DEAD 61. RIO TINTO
15. BRITISH AMERICAN TOBACCO 62. ROLLS-ROYCE HOLDINGS
16. BRITISH LAND 63. ROYAL BANK OF SCTL.GP.*
17. SKY 64. ROYAL DUTCH SHELL A(LON)
18. BT GROUP 65. ROYAL DUTCH SHELL B
19. BUNZL 66. RSA INSURANCE GROUP*
20. CABLE & WIRELESS COMMS. 67. SABMILLER 21. CADBURY DEAD - 08/03/10 68. SAGE GROUP
22. CAIRN ENERGY 69. SAINSBURY (J)
23. CAPITA 70. SCHRODERS*
24. CARNIVAL 71. SCHRODERS NV
25. CENTRICA 72. SSE
27. COMPASS GROUP 74. SHIRE
28. DIAGEO 75. SMITH & NEPHEW
29. EURASIAN NATRES.CORP. - 25/11/13 76. SMITHS GROUP
30. EXPERIAN 77. STANDARD CHARTERED*
31. FIRST GROUP 78. STANDARD LIFE*
32. FRIENDS PROVIDENT GROUP DEAD* 79. TATE & LYLE
33. G4S 80. TESCO
34. GLAXOSMITHKLINE 81. THOMAS COOK GROUP
35. HAMMERSON 82. THOMSON REUTERS DEAD
36. HOME RETAIL GROUP 83. TUI TRAVEL DEAD - 17/12/14 37. HSBC HDG. (ORD $0.50)* 84. TULLOW OIL
38. ICAP 85. UNILEVER (UK)
39. IMPERIAL TOBACCO GP. 86. UNITED UTILITIES GROUP
40. ICTL.HTLS.GP. 87. VEDANTA RESOURCES
41. INTERNATIONAL POWER - 02/07/12 88. VODAFONE GROUP
42. JOHNSON MATTHEY 89. WHITBREAD
43. KAZ MINERALS 90. WOLSELEY
44. KINGFISHER 91. WPP
45. LAND SECURITIES GROUP 92. XSTRATA DEAD - 03/05/13 46. LEGAL & GENERAL* 93. 3I GROUP
