University of Amsterdam
Faculty of Economics and Business
Bachelor Thesis
The impact of the UK short selling ban in 2008-2009 on
stock prices and volatility
Author: Supervisor:
Niels Fekkes (10366075) Ieva Sakalauskaite
Abstract:
On September 19, 2008 the Financial Services Authority in the UK imposed a short selling ban, which remains effective until January 16, 2009. This study examines the effect of the short selling ban on the cumulative abnormal return and the stock price return standard deviation of the banned financial companies. This study also analysis if there is an difference between the impact on global
systematically important financial companies and the other banned companies. The results suggest that the cumulative abnormal return during the short selling ban increased, but not significant. Secondly, during the short selling ban the cumulative abnormal return of the global systematically important financial companies decreased significantly. Finally, due to the ban the volatility increased
significantly, but there was not a significant difference between the short selling ban impact on the volatility of global systematically important financial companies and the other banned companies. Keywords: short selling, ban, volatility, cumulative abnormal return, FSA
JEL classification: G01, G14, G18
2
Statement of Originality
This document is written by Niels Fekkes who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is 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.
3 1. Introduction
Short selling allows people, who expect price declines, to participate in the market. With short selling an investors borrows a stock from a broker and sells the stock for the current stock price. For this service the broker requires a premium. The investor intends to buy back the stock later for a lower price. Short selling gives the investor the opportunity to make a profit, when the stock price decreases. That is why investors use short selling to speculate on price declines. Short selling is also used as a hedging tool or an arbitrage tool. Under normal conditions short selling contributes to more efficient stock prices, but in times of financial distress short selling will cause price declines of the financial companies securities.
Short-selling became particularly relevant after the start of the financial crisis, when several financial companies faced problems. On September 15, 2008 Lehman Brothers went bankrupt. The bankruptcy influenced not only the stock market in the US, but it affected the entire world. Stock markets all over the world had to deal with declines in stock prices. Especially financial companies were affected by the downward pressure on their stock prices. As a result, investors could speculate on price declines using short sales.
As short selling will cause an even higher downward pressure on stock prices, regulators in different countries implemented constraints on short selling to prevent collapses of other financial companies. One of the countries that has implemented short selling measures was the US. The US took temporary measures after the collapse of Lehman Brothers. The Securities and Exchange Commission introduced short selling restrictions for covered short selling as well as naked short selling. The short selling restrictions were effective for almost eight hundred financial sector companies. With the short selling ban the Securities and Exchange Commission wanted to protect the integrity and quality of the securities market and strengthen the investor’s confidence.
Also in the UK short selling restrictions were imposed after the collapse of Lehman Brothers. Hector Sants, chief executive of the Financial Services Authority, said the following about the decision to implement short selling restrictions: "While we still regard short-selling as a legitimate investment technique in normal market conditions, the current extreme circumstances have given rise to disorderly markets. As a result, we have taken this decisive action, after careful consideration, to protect the fundamental integrity and quality of markets and to guard against further instability in the financial sector." The FSA imposed a short selling ban for covered and naked short sales of financial sector companies. The short selling ban was announced on September 18, 2008 and was effective from September 19, 2008 until January 16, 2009. Thirty days after the implementation of the short selling restrictions the policy measurement would be reviewed.
4 After the implementation of the recent short selling constraints, questions arose about the impact of those measures. Based on the existing literature about short selling, there is some discussion about the impact of short selling constraints. Miller (1977) pointed out that restrictions on short selling, will cause inefficient stock prices. This is because prices do not reflect the beliefs of the less optimistic investors. As a consequence, stocks are overpriced. On the other hand, Diamond and Verrecchia (1987) argue that short selling restrictions do not bias prices upwards, but short selling constraints only affects the speed of adjustment of security prices. Helmes, Henker and Henker (2010) did their
research about the impact of a short selling ban on the volatility of the stock prices. They conclude that a short selling ban will increase the volatility of the stock prices. This study tries to answer the
following research question:
What is the impact of the 2008-2009 UK short selling ban on the stock prices and volatility of the banned financial institutions?
In order to test the hypothesis about the overpricing of banned stocks, the cumulative abnormal returns will be computed. After that a regression will follow to measure the impact of the short selling ban on the cumulative abnormal returns. The volatility will be measured by calculating the standard deviation of the stock price return. The standard deviation and cumulative abnormal return are calculated during the period before the ban, during the ban and after the short selling ban. This paper also makes a distinction between the impact on global systematically important financial institutions and the impact on other banned financial institutions.
The remainder of this paper is structured as follow. Section 2 describes some general information about short selling, the purpose of short selling, the UK short selling ban and the global systematically important financial institutions. Followed by a review of the related literature. Section 3 provides the hypotheses, based on the related literature. In order to test these hypotheses, this section also describes the data and the methodology that is used. Section 4 presents the results of the methodology. Finally, section 5 provides the conclusion of this paper.
5 2. Literature review
2.1 General information about short selling
Short selling gives an investor the opportunity to make a profit, when a stock price decreases. There are two types of short selling. The first one is called covered short selling. With a covered short sale an investor borrows a share from a broker and sells it right away. After a while the investor buys a share of the same stock in order to give it back to the lender. So the investor can make a profit by buying the securities for a lower price, but the investor has to pay a premium to the lender. The total profit of the short seller is the decline of the stock price minus the premium paid to the lender (Bodie, Kane, & Marcus, 2014). A more controversial way of short selling is naked short selling. With naked short sales, the trader does not borrow the stock or ensures it to borrow it later on. In some countries naked short sales are not allowed (Helmes, Henker, & Henker, 2010). In both ways short selling allows selling a particular kind of security, that one does not own.
2.2 Purposes of short selling
According to Staley (1997), short selling has several purposes. First of all, an investor can use short selling to speculate on price declines. Investors speculate on the fact that a particular kind of security is overpriced. They expect that in the future the price of the security will decrease. In this way investors can make a profit when the price of a security declines. Another reason why short selling is used, is that short selling can be used as a hedging technique. Short sales allow an investor to hedge against the market risk. Finally, short selling is used by arbitrageurs. Arbitrageurs buy one security and sell another one. In this way arbitrageurs can make a profit. Levy and Ritov (2001) wrote their paper about the importance of short selling in portfolio optimization. In a portfolio with many assets short selling can increase the return per unit of risk. Figure 1 plots the relation between the Sharpe ratio and the number of stocks. With short sales the Sharpe ratio will increase.
6
Figure 1: Sharpe ratio with and without short selling1
2.3 UK short selling constraints
After the start of the financial crisis, financial companies faced liquidity problems. The failure of Lehman brothers in September 2008 reduced investors’ confidence in the stability of the financial sector in the US and other developed countries. The failure of Lehman Brothers had not only effect on the financial sector in the US, but it also affected stock markets across the world. To prevent collapses of other financial companies regulators in different countries implemented several measures. These measures include constraints and bans on short selling. In the UK the FSA imposed a short selling ban for covered and naked short sales of financial sector companies. According to Hector Saints, the short selling ban must protect the fundamental integrity and quality of markets and to guard against further instability in the financial sector. The short selling ban was announced on September 18, 2008 and was effective on September 19, 2008 until January 16, 2009. Thirty days after the implementation of the short selling restrictions the policy measurement would be reviewed.
Woolridge and Dickinson (1994) did their empirical research about the effect of short selling on stock prices. They based their conclusion on an analysis of overall market data and of individual institutions traded on the New York Stock Exchange, Amex and OTC Markets. Their study covers the 1986 until 1991 period. They found evidence that short selling does not lead to lower stock prices. Another
1
Levy, M., & Ritov, Y. (2001). Portfolio Optimization with Many Assets: The Importance of
Short-Selling. Working paper. Retrieved from website eScholarship University of California:
7 conclusion they make in their research, is that investors, who are active in short selling, do not earn abnormally high or low returns. Their study denies that short sellers earn abnormal returns at the cost of less informed traders by decreasing stock prices using short selling. So Woolridge and Dickinson reject the hypothesis, which state that short selling will cause an even higher downward pressure on stock prices. So why imposing short selling measures.
The theoretical article about short selling of Brunnermeier and Oehmke (2014) provides a rationale for introducing short selling bans. They wrote their paper about predatory short selling. Predatory short selling occurs when financial institutions face liquidity problem or financial weakness. Investors want to exploit this. In their theoretical model, predatory short selling occurs in equilibrium, because the decrease in equity valuation caused by short sellers leads to withdraw funding from financial companies by non-insured depositors and short term creditors. Therefore short selling can force the financial firms to liquidate long-term asset holdings at a discount. The decrease in value can allow short sellers to break even their positions. Their analysis shows us that financial companies can be vulnerable to attacks from predatory short selling when their balance sheets are weak. Short selling can compel a complete liquidation of the financial firm’s long term asset holdings. This probability increases when short sellers are better able to coordinate their actions. The findings of Brunnermeier and Oehmke (2014) can justify the short selling constraints of financial institutions during financial distress.
2.4 Global Systematic Important
Financial Stability Board published an integrated set of policy measures to assign the systemic and moral hazard risks associated with systemically important financial institutions. In that publication, the Financial Stability Board identified an initial group of global systematically financial institutions. The Financial Stability Board made a distinction between Global Systematically Important Banks and Global Systematically Important Insurers. Both groups will be updated annually based on new data and published by the Financial Stability Board each November. According the most recent list of Global Systematically Important Banks, HSBC, Barclays, Royal Bank of Scotland and Standard Chartered are defined as global systematically important. Aviva and Prudential are identified as Global Systematically Important Insurers.
2.5 Related literature
This section will introduce the existing literature about the effects of short selling constraints on financial markets. First, it will discuss the theoretical predictions. Then, empirical evidence will be provided.
8
2.4.1 Theoretical predictions on the effects of short selling restrictions
Under normal conditions short selling contributes to more efficient stock prices. In times of financial distress short selling will cause price declines in the securities of financial companies. Therefore short selling comes under attack during times of financial distress.
However, there is an ongoing debate about the effects of short selling restrictions. There are several scientific studies written about covered short selling, naked short selling and the effect of restrictions on going short. First of all Miller (1977) concluded that short selling restrictions will cause inefficient stock prices. In a market that is efficient, all available information is right away priced in. When people, who believe that asset prices will decrease, cannot short sell, this information is not
incorporated in asset prices. This will result in inefficient stock prices. As a result, securities will be overpriced, because a small group of optimistic investors can drive up the prices of securities. When short selling is limited, the securities market’s demand is only determined by optimistic investors (Miller, 1977).
On the other hand, Diamond and Verrecchia (1987) argue that short selling restrictions do not bias prices upwards, because traders have rational expectations. However, short selling constraints will influence the reflection of information on prices. Diamond and Verrecchia (1987) used a theoretical model to measure the impact of short selling constraints on security prices and the speed of
adjustment. By introducing measures, which limit short trading, policy makers influence the speed of adjustment of the stock prices. This is especially the case in times of bad news. In times of bad news, investors have less optimistic beliefs. During these times some investors expect a decrease in stock prices. When short selling is allowed, they can earn profits with decreasing stock prices. If short selling is not allowed security prices do not reflect those beliefs. So adjusting security prices takes more time, but at the end prices will also reflect the beliefs of the less optimistic investors. The conclusion of Diamond and Verrecchia (1987) differs from the conclusion of Miller (1977), because Miller beliefs that short selling constraints bias stock prices upwards. Diamond and Verrecchia reject the conclusion of Miller, because investors have rational expectations and concluded that short selling constraints only have effect on the speed of adjustment of stock prices.
Bai, Chang and Wang (2006) did their research about the impact of short selling constraints on the volatility of stock prices. They assumed a fully rational expectations equilibrium model. In such kind of model investors trade to share risk and speculate on private information. Short selling restrictions reduce both types of trades. Therefore short selling constraints reduce the allocation and informational market efficiency. Limiting the two types of trades has opposite impact on security prices. When investors trade to share risk, limiting short sales shifts the demand for the security upwards. Therefore the prices of those securities will increase. On the other hand, limiting short sales, driven by
9 uncertainty about the security as perceived by the less informed investors. Therefore the demand of the security will decrease and so does its price. Bai, Chang and Wang (2006) also concluded that in a less efficient market the volatility of the stock prices will increase, because prices fail to be informative. Based on the theoretical papers about the impact of short selling constraints on stock prices, there is an ongoing debate. Miller (1977) concluded that short selling constraints bias prices upwards. On the other hand, Diamond and Verrecchia (1987) concluded that short selling restrictions do not bias prices upwards. Bai, Chang and Wang (2006) argued that short selling constraints could bias prices upwards, but also downwards. So there is not a clear answer what the effect is of a short selling ban on stock prices. Based on the study of Bai, Chang and Wang (2006), the expectation is that the volatility of the stock price return will increase due to a short selling ban.
2.4.2 Existing empirical evidence on the effects of short selling restrictions
Boehmer and Wu (2013) did their empirical research about short selling and price discovery process. Their research is based on 1361 stocks, traded on the New York Stock exchange. The period, which is covered in this study, starts in 2005 until 2007. Price discovery is the process where new information is reflected in security prices. They found evidence that when there are active short traders, the process of price discovery improves. With short selling prices appear to be closer to efficient values. This is in line with the theoretical evidence of Diamond and Verrecchia (1987), who concluded that a short selling ban has a negative influence on the price discovery process.
Saffi and Sigurdsson (2011) concluded that short selling restrictions have impact on the stock price efficiency. Their research includes more than 12600 stock from twenty-six countries between 2005 and 2008. They used the equity lending supply of stocks in their research. In times of constraints on short selling there is no equity lending supply. When there is no lending supply, prices of stocks are more inefficient, because a higher level of equity lending supply increases the speed with which information is reflected into prices. This confirms the conclusion of Miller (1977). Saffi and
Sigurdsson also found evidence, that allowing short selling does not lead to less stable security prices or abnormal negative returns. Their evidence suggest that short selling constraints do not achieve the desired goal of stabilizing security prices.
There are also studies done about the impact of short selling constraints on the volatility of stock prices. Kraus and Rubin (2003) focused on the initiation of index option in Israel. The initiation of index option in Israel is consistent with the fact that, there were short sale restrictions when index options were introduced. They found evidence that this caused higher price volatility. They based their conclusion on an empirical model to see if the volatility changes. They measured the volatility of an Israeli index called The MAOF, which is an capped index weighted by the market capitalization of the twenty-five largest institutions of the Israeli economy. Charoenrook and Daouk (2005) support the
10 fact that short sale restrictions increase the volatility of the return of stocks. Their empirical study is based on the historical short selling regulations of 111 countries. So when short selling is allowed the volatility of the stock prices is lower. They also conclude that constraints on short selling bias stock prices upwards, but short sale restrictions have no effects on the skewness of the stock returns or the probability of a market crash. However, they concluded that allowing short selling will improve the stock market quality. So both empirical papers confirms the conclusion of Bai, Chang and Wang (2006), who concluded that the volatility will increase due to short selling restrictions.
There are also scientific articles written about the short selling restrictions imposed in countries after the collapse of Lehman Brothers. Beber and Pagano (2013) measures the effect of the short selling bans on the market liquidity, the price discovery process and stock overpricing. They use panel and matching techniques to investigate the effects. Their research also included the UK. Their overall conclusion was that short selling bans were detrimental for liquidity. This was especially the case for stocks with small capitalization and no listed options. Their results of the effect on price discovery confirmed the conclusion of Boehmer and Wu (2013). Beber and Pagano (2013) found evidence that short selling constraints reduce price discovery, especially in bear markets. Finally, short selling bans failed to support prices, except for the US financial stocks.
Helmes, Henker and Henker (2010) did their research about the impact of the ban on short selling the market quality. The focus on this study was on the Australian shorting ban. Australian regulators imposed a short selling ban in 2008 until May 2009. They measured market quality as trading activity, bid-ask spreads and volatility. Differences are examined and compared over the period before the ban, during the ban and after the ban. Their study found reduced trading activity, increased bid and ask spreads and increased volatility. As a result of that, they conclude that there is strong evidence that the short selling ban in Australia has effect on the market quality in a negative way.
Frino, Lecce and Lepone (2011) also focused on the short selling bans from 2008. They want to measure the impact of the short selling bans on the market quality by using the same measures for market quality as Helmes, Henker and Henker (2010) did. Frino, Lecce and Lepone based their research on short selling bans of different countries. One of them was the UK short selling ban. Market quality is measured by using bid-ask spreads, price volatility and trading activity. They concluded that the market quality declines during short selling constraints. They came to this conclusion, because of wider bid-ask spreads, increased price volatility and reduced trading activity. They also measure the impact of the bans on the stock prices, by computing the abnormal returns. They calculated the cumulative abnormal returns 10 days before and after the implementation of short selling ban. They found evidence that during the ban the abnormal returns are significantly higher. This is consistent with what Miller (1977) already pointed out. Frino, Lecce and Lepone measured that
11 the cumulative abnormal return after the implementation of the 2008-2009 UK short selling ban was 5.43%.
There were also studies done about the impact of the short selling ban in the US. One of the studies was the one of Boehmer, Jones and Zhang (2013). In September 2008 the Securities and Exchange Commission imposed a short selling ban for almost thousand stocks. Their study examines the effect on market quality, shorting activity, the aggressiveness of short sellers and stock prices. For this research the impact on market quality and stock prices are important. They found that during the short selling ban the market quality worsens. They measured market quality by using three variables: effective spreads, price impact and realized spreads. The negative effect on market quality confirms the conclusion of Helmes, Henker and Henker (2010), but they used bid-ask spreads, price volatility and trading activity as measures of market quality. Boehmer, Jones and Zhang also found evidence that the banned stock prices increased. Only, they noticed that the increase of the banned stock prices can also be the result of the bailout program.
Another study which is done about the short selling ban imposed by the Securities and Exchange Commission is the one of Boulton and Braga-Alves (2010). To test the hypothesis that during a short selling ban stocks are overpriced, they calculated the abnormal returns of the restricted firms. They compare the abnormal returns of the restricted institutions with the abnormal returns of a matched sample of unrestricted institutions. Their conclusion was in line with Miller (1977). They found evidence that banned stock were overpriced. Boulton and Braga-Alves also investigated the market quality. They did this by analysing bid-ask spreads, trading volume and stock return volatility. Their conclusion was that short selling constraints have a negative effect on the market quality. In this case the volatility of the stock return is interesting. They measured volatility in two ways. First they measured volatility using the standard deviation of the daily returns. Second they use tick-by-tick quote data to calculate intraday volatility as the average standard deviation of five minute returns computed using bid-ask midpoints measured at five minute intervals during normal trading hours. During the ban the volatility increased, but not significantly more than the matched sample of unrestricted institutions.
Marsh and Payne (2012) did their research about the impact of the UK short selling ban on the market quality. They use trading activity, liquidity, price efficiency and price discovery, as measures of market quality. They measure the impact on price efficiency by calculating the proportion of variation in returns that is driven by information, as opposed to noise. They found that stock trading volume of financial companies decreased during the ban. Also they found evidence that the liquidity decreased. Finally, they found that price efficiency as well as the price discovery declines during the short selling ban. They concluded that due to the impact on these variables the market quality in the UK during the ban decreased, which was not in line with the aim of the short selling ban.
12 This paper will use alternative measures of market quality from Marsh and Payne to assess the effects of the short selling ban in the UK. First of all, the effects on market efficiency will be evaluated by comparing the cumulative abnormal returns of the banned stocks before, during and after the ban. Rather than calculating the proportion of variation in returns that is driven by information opposed to noise, as in Marsh and Payne. Another way to measure the market quality is the volatility. The volatility will be computed as the standard deviation of the daily stock price return during the pre ban period, the ban period and the post ban period. Marsh and Payne did not use volatility, as measure of market quality. Also this paper will focus on the difference of the short selling ban impact between global systematically important financial institutions and the other banned financial companies.
13 3. Data and methodology
This section provides the hypothesis, based on the related literature. This section also describes the data and methodology that will be used in this thesis.
3.1 Hypothesis development
The first hypothesis concerns the effects of the short selling ban on stock prices. According to Miller (1977) short selling restrictions will cause overpriced stocks, because less optimistic investors cannot participate in the market. Overpriced securities must lead to higher abnormal returns. Therefore the first hypothesis will be:
During the short selling ban, banned stocks will have higher cumulative abnormal returns than when short selling is allowed. This effect will be lower for global systematically important financial institutions.
Also the impact of short selling restrictions on the volatility of the banned stock prices will be tested. According to the above literature about the impact of the limitations of short selling, a ban will increase the volatility of the banned stocks. Therefore the second hypothesis will be:
During the short selling ban, the standard deviation of the banned stock price return is higher than when short selling is allowed. This effect will be lower for global systematically important financial institutions.
3.2 Methodology
This section describes the methodology that will be used to measure the impact of the short selling ban on the financial market quality and financial stability measured as the cumulative abnormal return and the volatility.
After calculating the cumulative abnormal returns and stock volatility for all the financial institutions, which were banned, the following model will be estimated.
𝑌𝑡 = 𝛼 + 𝛽1𝑋𝑏𝑎𝑛,𝑡+ 𝛽2𝑋𝐺𝑆𝐼+ 𝛽3𝑋𝑏𝑎𝑛,𝑡𝑋𝐺𝑆𝐼+ 𝛽4𝑋𝑝𝑟𝑒 𝑏𝑎𝑛,𝑡+ 𝛽5𝑋𝑝𝑟𝑒 𝑏𝑎𝑛,𝑡𝑋𝐺𝑆𝐼 + 𝛽6𝑋ln(𝑚𝑎𝑟𝑘𝑒𝑡 𝑣𝑎𝑙𝑢𝑒,𝑖),𝑡+ 𝛽7𝑋ln(𝑡𝑟𝑎𝑑𝑒 𝑣𝑜𝑙𝑢𝑚𝑒,𝑖),𝑡+ 𝜀𝑡
Where 𝑌𝑡 is the dependent variable cumulative abnormal return or stock volatility. For both
regressions the heteroscedasticity is tested. In order to test for heteroscedasticity a Breusch-Pagan test is performed. When there is heteroscedasticity, the estimated variance of the residuals from the regressions depends on the values of the independent variables. According to the Breusch-Pagan test, the residuals are heteroscedastic. Therefore the model is corrected for heteroscedasticity by using robust standard errors.
14 The first independent variable 𝑋𝑏𝑎𝑛,𝑡 is a dummy variable, which has the value one, if the cumulative abnormal return or stock volatility is measured during the short selling ban. This variable is important to test the first hypothesis. The expectation is that the estimation of 𝛽1 is significant positive to measure the cumulative abnormal returns, as well as the standard deviation of the daily stock price return.
The second variable 𝑋𝐺𝑆𝐼 is also a dummy variable, which have the value one, if the financial
institution is a global systematically important bank or a global systematically important insurer. The variable, global systematically important bank or global systematically important insurer, is included in the regression, because this study expects that it will influence the stock prices of these financial institutions during financial distress. Therefore it will also influence the abnormal return. The expectation is that the cumulative abnormal returns of the global systematically important financial institutions will be closer to zero than the ones of the other banned financial companies. Also the standard deviation of the daily stock price return is expected to be lower for a global systematically important bank or a global systematically important insurer, than for the other financial institutions. The model includes also a interaction variable 𝑋𝑏𝑎𝑛,𝑡𝑋𝐺𝑆𝐼 between the dummy variable of the ban and the dummy variable of the global systematically important financial institutions. The interaction variable measures the difference of the short selling ban impact between global systematically important financial institutions and the one which are not global systematically important. The expectation is that the estimation of 𝛽3 will be negative, because the expectation is that the impact of the short selling ban on the cumulative abnormal returns and volatility for global systematically important financial institutions will be lower.
The independent control variable 𝑋𝑝𝑟𝑒 𝑏𝑎𝑛,𝑡 is a dummy variable, which has the value one, if the cumulative abnormal return or stock volatility is measured during the pre ban period. This variable is included in the model, because it makes a difference between measuring the cumulative abnormal return or stock volatility before and after the short selling ban. Therefore there is also an independent control variable 𝑋𝑝𝑟𝑒 𝑏𝑎𝑛,𝑡𝑋𝐺𝑆𝐼, which is an interaction variable between the pre ban variable and the variable of the global systematically important financial companies.
The sixth and seventh variables of the model are also control variables. The sixth variable
𝑋ln(𝑚𝑎𝑟𝑘𝑒𝑡 𝑣𝑎𝑙𝑢𝑒,𝑖),𝑡 is the natural logarithm of the banned financial companies average market value
during the pre ban period, the ban period and the post ban period. The market value is computed by multiplying the share price by the number of ordinary shares in issue. The amount in issue is updated, when new tranches of stock are issued or after a capital change. Finally the seventh variable
15 volume during the pre ban period, the ban period and post ban period. The daily trading volume is measured by the number of shares traded for a stock on a particular day.
3.3 Data
This empirical analysis will use information on stock prices affected by the short selling ban in the UK during May 23, 2008 to May 18, 2009. This sample period was chosen to compare market efficiency during periods of equal length before, during and after the ban. As the short selling ban lasted for 83 trading days, introduced on September 18, 2008 and was lifted on January 16, 2009, the resulting three periods are as follow:
1. The time period before the short selling ban will be May 23, 2008 until September 18, 2009, called the pre ban period.
2. The time period during the short selling ban will be September 19, 2008 until January 16, 2009, called the ban period.
3. The time period after the short selling ban will be January 17, 2009 until May 18, 2009, called the post ban period.
Although the FSA banned covered and naked short selling of 34 financial institutions, this analysis uses information on 22 of them. Nine of the 34 financial firms, which were banned, were too small to take account for. They were traded on the FTSE Small Cap or the Alternative Investment Markets rather than the FTSE 350 which is the source of data in this analysis. Three banned stocks were dropped out of the sample, because of incomplete data due to acquisition and merger activity. Table 1 provides a list of the banned financial institutions, which were used in this analysis.
Table 1: Banned financial institutions
1. Aberdeen Asset Management PLC 12. Lloyds TSB Group PLC
2. Admiral Group PLC 13. Old Mutual PLC
3. Alliance Trust PLC 14. Provident Financial LTD
4. Aviva PLC 15. Prudential PLC
5. Barclays PLC 16. Rathbone Brothers
6. Brit Insurance Holdings PLC 17. Royal Bank of Scotland Group PLC 7. Close Brothers Group PLC 18. RSA Insurance Group PLC
8. Friends Provident PLC 19. Schroders PLC
9. HSBC Holdings PLC 20. St James’s Place PLC
10. Investec PLC 21. Standard Chartered PLC
16 I use the daily stock prices of the twenty-two banned financial companies to calculate the stock price return and the volatility. The volatility will be measured as the standard deviation of the daily stock prices return. To measure abnormal returns I use stock prices of the 22 banned companies and the FTSE 350 market index. The interest rate on a ten year US Treasury bond is used as the risk free rate. The benchmark this study will use to compute the daily abnormal returns will be the Capital Asset Pricing Model. The Capital Asset Pricing Model provides the following formula:
𝐸(𝑅)𝑖,𝑡− 𝑅𝑓 = 𝛼𝑖+ 𝛽𝑖∗ (𝑅𝑀𝑘𝑡,𝑡− 𝑅𝑓,𝑡) + 𝜀𝑖,𝑡
Where 𝐸(𝑅)𝑖,𝑡, the daily expected return of stock i is. In this equitation the daily expected return of the stock is the dependent variable. The daily market return, 𝑅𝑀𝑘𝑡, is the independent variable. The market index return is determined by the FTSE 350. On both sides of the equation the daily risk free rate, 𝑅𝑓
, is subtracted. This study assumes that a year will consist of 252 trading days. The risk free
rate is determined by the interest rate on a US treasury bond, because of the AAA credit rating of US treasury bonds. Using Ordinary Least Squares the parameters 𝛼𝑖 and 𝛽𝑖,𝑡are calculated. The securities
𝛽𝑖,𝑡 measures the sensitivity of the daily stock return to the excess daily market return and 𝛼𝑖 is the intercept coefficient. To estimate the parameters, 𝛽𝑖,𝑡 and 𝛼𝑖, the period will start three years before the sample period. Therefore the period starts May 22, 2005 until May 22, 2008.Secondly, the parameters are used to calculate the daily abnormal returns of each financial institution. The daily abnormal returns are computed using the following formula.
𝐴𝑅𝑖,𝑡= 𝑅𝑖,𝑡− 𝑅𝑓,𝑡− (𝛼𝑖+ 𝛽𝑖∗ (𝑅𝑚𝑘𝑡,𝑡− 𝑅𝑓,𝑡))
The daily abnormal return is the difference between the actual daily return minus the expected daily return. The daily abnormal returns will be calculated in the pre ban period, the ban period and the post ban period.
In order to estimate the model, which is used in this study, the cumulative abnormal returns for the banned financial companies has to be computed. The cumulative abnormal returns are calculated using the following formula.
𝐶𝐴𝑅𝑖,𝑡= ∑ 𝐴𝑅𝑖,𝑡
The cumulative abnormal returns will be computed for all the banned financial institutions for the pre ban period, the ban period and the post ban period.
The daily stock prices, risk free rates and the FTSE 350 market index are retrieved from the
17 Systematically important Insurers are retrieved from the Financial Stability Board website. The
regression model also includes control variables, namely the financial institutions market value and trading volume. These data are retrieved from the DataStream database.
Table 2 provides the descriptive statistics of the used variables during the pre ban period, the ban period and the post ban period for the banned companies. Table 2 shows that during the short selling ban the mean of the cumulative abnormal return decreases, followed by an increase of the cumulative abnormal return mean during the post ban period. According to the descriptive statistics of the volatility, the average standard deviation of the stock price return during the short selling ban increased by 2,7557%. Table 2 also shows that the average natural logarithm of the market value decreased during the ban period and the post ban period. Finally, from table 2 we can conclude that during the short selling ban the average natural logarithm of the trading volume is lower than during the pre and post ban period. Table 3 provides the correlation matrixes. The correlation matrixes show that there exist no perfect multicollinearity between the used variables.
Table 2a, b, c: Descriptive statistics
a) Pre ban period
b) Ban period
c) Post ban period
Mean Median Std. Dev. Min Max Skewness
CAR 0,003123 0,015552 0,178668 -0,37243 0,259922 -0,177143 Volatility 0,033091 0,033025 0,007028 0,017369 0,04547 -0,43255 ln(mkt value) 8,329617 8,082075 1,489578 5,944676 11,50556 0,3808705 ln(trading vol) 8,606295 8,802889 1,888407 4,380278 11,72289 -0,231494
Mean Median Std. Dev. Min Max Skewness
CAR -0,14377 -0,09477 0,322447 -1,06965 0,19499 -1,653355 Volatility 0,060648 0,060752 0,016628 0,03478 0,093017 0,2788476 ln(mkt value) 8,115521 7,925352 1,382976 5,883887 11,3942 0,492676 ln(trading vol) 8,156159 8,393417 1,952604 3,8193 11,35376 -0,274503
Mean Median Std. Dev. Min Max Skewness
CAR 0,029656 -0,03859 0,408167 -0,69767 1,226394 1,1892349 Volatility 0,054074 0,050803 0,031275 0,018891 0,121712 1,0750862 ln(mkt value) 8,022679 7,855157 1,314647 5,791992 11,1504 0,5104288 ln(trading vol) 8,439859 8,396865 2,103621 3,953697 11,94589 -0,122344
18
Table 3a, b: Correlation matrix
a) Correlation matrix including the cumulative abnormal return
b) Correlation matrix including the standard deviation
CAR Ban GSI Ban*GSI Pre Pre*GSI ln(mkt value) ln(trade vol.) CAR 1,000000 Ban -0,236322 1,000000 GSI 0,033701 -0,000000 1,000000 Ban*GSI -0,378648 0,447214 0,516398 1,000000 Pre 0,088798 -0,500000 -0,000000 -0,223607 1,000000 Pre*GSI -0,024184 -0,223607 0,516398 -0,100000 0,447214 1,000000 ln(mkt value) 0,013587 -0,020839 0,772911 0,386305 0,089544 0,471557 1,000000 ln(trade vol.) 0,054005 -0,088843 0,522757 0,229995 0,074646 0,274171 0,821296 1,000000
Std. Dev. Ban GSI Ban*GSI Pre Pre*GSI ln(mkt value) ln(trade vol.) Std. Dev. 1,000000 Ban 0,342100 1,000000 GSI 0,348121 -0,000000 1,000000 Ban*GSI 0,323958 0,447214 0,516398 1,000000 Pre -0,486515 -0,500000 -0,000000 -0,223607 1,000000 Pre*GSI -0,193322 -0,223607 0,516398 -0,100000 0,447214 1,000000 ln(mkt value) 0,324051 -0,020839 0,772911 0,386305 0,089544 0,471557 1,000000 ln(trade vol.) 0,481780 -0,088843 0,522757 0,229995 0,074646 0,274171 0,821296 1,000000
19 4. Results
This section discus the results of the cumulative abnormal return regression and the regression on the volatility. First the regression results of the cumulative abnormal return will be discussed and after that the regression results of the standard deviation will follow. Finally, an interpretation of the results are given.
4.1 Regression results of the cumulative abnormal return
Table 4 provides the regression result with the cumulative abnormal return, as the dependent variable. The estimation of the 𝑋𝑏𝑎𝑛,𝑡 coefficient is 0,0771. This means that during the short selling ban the average cumulative abnormal returns of the banned financial companies increased with 7,71%, holding other variables constant. This is in line with the expectation that during short selling
restrictions the cumulative abnormal returns will increase for banned companies. Only the estimation of the 𝑋𝑏𝑎𝑛,𝑡 coefficient is not significant, which is not in line with the first hypothesis.
Table 4 also show us the estimation of the 𝑋𝐺𝑆𝐼 coefficient. The estimated value in the model is 0,5383. This implies that the cumulative abnormal returns is positively affected by global systematically important financial institutions. The cumulative abnormal return of a global systematically important financial company is 53,83% higher than the other financial companies, holding everything else constant. The estimation of the 𝑋𝐺𝑆𝐼 coefficient is significant at a 1% level. This was not in line with the expectation that the cumulative abnormal return for global systematically important financial companies will be closer to zero.
Finally, table 4 provides the estimation of the 𝑋𝑏𝑎𝑛,𝑡𝑋𝐺𝑆𝐼 interaction coefficient. During the short selling ban the cumulative abnormal return of the global systematically important financial institutions is 38,17% lower than for the other financial institutions. The estimation of the 𝑋𝑏𝑎𝑛,𝑡𝑋𝐺𝑆𝐼 interaction coefficient is significant at a 1% level and suggest that during the short selling ban the cumulative abnormal return of a global systematically important financial company is 92,01% lower. The expectation was that the increase of the cumulative abnormal return of a global systematically
important company caused by the short selling ban was less than for the other banned companies. The results were not in line with the expectation, because the short selling ban caused a lower cumulative abnormal return of a global systematically important company than during the pre and post ban period.
20
Table 4: Regression output of the cumulative abnormal return
Variable Estimated coefficient
Ban 0.0771 (0.0894) GSI 0.5383** (0.2476) Ban*GSI -0.9201*** (0.2605) Pre 0.1449 (0.0875) Pre*GSI -0.6303*** (0.1959) Ln(Mktvalue) 0.0019 (0.6023) Ln(Tradevol) -0.0008 (0.0327) Intercept -0.1255 Oberservations 66 R squared 0.3445
Table 4 provides the regression results with the cumulative abnormal return as dependent variable. The variable Ban equals 1 if the cumulative abnormal return is measured during the short selling ban. The variable GSI equals 1 if the company is global systematically important. The interaction variable Ban*GSI equals 1 if the cumulative abnormal return of a global systematically important financial institution is measured during the short selling ban. The variable Pre equals 1 if the cumulative abnormal return is measured during the pre ban period. The interaction variable Pre*GSI equals 1 if the cumulative abnormal return of a global systematically important financial institution is measured during the pre ban period. Ln(Mktvalue) is the natural logarithm of the market value. Ln(Tradevol) is the natural logarithm of the trading volume.
* = significant at the 10 percent level ** = significant at the 5 percent level *** = significant at the 1 percent level
21
4.2 Regression results of the standard deviation
Table 5 provides the regression results with the standard deviation of the stock price return, as
dependent variable. First of all, table 5 shows the estimation of the 𝑋𝑏𝑎𝑛,𝑡 coefficient, which is 0.0139. According to the estimation of the 𝑋𝑏𝑎𝑛,𝑡 coefficient, this means that during the short selling ban the standard deviation of the banned stock return is 1,39% higher, holding other variables constant. The estimation of the 𝑋𝑏𝑎𝑛,𝑡 coefficient is significant at a 5% level. The results confirms the hypothesis that during a short selling ban the standard deviation will increase significantly.
Table 5 shows also the estimation of the 𝑋𝐺𝑆𝐼 coefficient, which is 0,0322. This coefficient would imply that the standard deviation of the stock price return for global systematically important financial institutions is 3,22% higher than for the other financial institutions. This result is significant at a 1% level. The expectation was that the standard deviation for global systematically important financial companies is lower than the standard deviation of the other financial companies. The regression results reject this expectation.
Finally, the expectation was that during the short selling ban the standard deviation of the banned stock price return is higher than when short selling is allowed. This effect will be lower for global systematically important financial institutions. The coefficient of the interaction variable 𝑋𝑏𝑎𝑛,𝑡𝑋𝐺𝑆𝐼 confirms this expectation. The coefficient of the variable 𝑋𝑏𝑎𝑛,𝑡𝑋𝐺𝑆𝐼 is -0,0139, which means that during the short selling ban the standard deviation of the stock price return for global systematically important financial institutions is 1,39% lower. So the effect of the short selling ban is less for global systematically important financial institutions than for the other financial institutions, but the
22
Table 5: Regression output of the standard deviation
Variable Estimated coefficient
Ban 0.0139** (0.0056) GSI 0.0322** (0.0135) Ban*GSI -0.0139 (0.0134) Pre -0.0129** (0.0055) Pre*GSI -0.0245*** (0.0123) Ln(Mktvalue) -0.0095** (0.0039) Ln(Tradevol) 0.0095** (0.0021) Intercept 0.0418 Oberservations 66 R squared 0.6352
Table 5 provides the regression results with the standard deviation as dependent variable. The variable Ban equals 1 if the standard deviation is measured during the short selling ban. The variable GSI equals 1 if the company is global systematically important. The interaction variable Ban*GSI equals 1 if the standard deviation of a global systematically important financial institution is measured during the short selling ban. The variable Pre equals 1 if the standard deviation is measured during the pre ban period. The interaction variable Pre*GSI equals 1 if the standard deviation of a global systematically important financial institution is measured during the pre ban period. Ln(Mktvalue) is the natural logarithm of the market value. Ln(Tradevol) is the natural logarithm of the trading volume.
* significant at the 10 percent level ** significant at the 5 percent level *** significant at the 1 percent level
23
4.3 Interpretation results
According to the regression results of the cumulative abnormal returns, the short selling ban has not an significant impact on the cumulative abnormal returns. This reject the conclusion of Frino, Lecce and Lepone (2011). They found a cumulative abnormal return in the UK, which was significant higher due to the short selling ban. This can be a possible explanation for the theory of Diamond and Verrecchia (1987), who concluded that at the end short selling restrictions do not bias stock prices upwards. However, short selling restrictions do influence the price discovery process. Based on the results of this paper, the conclusion is that short selling restrictions do not bias prices upwards, because the short selling ban does not have a significant impact on the cumulative abnormal returns. Only according the results of Frino, Lecce and Lepone (2011), the stock prices are affected by the short selling ban, because they measure a significant cumulative abnormal return during 10 days after the
implementation of the short selling ban. This suggest that the UK short selling ban has an negative influence on the price discovery process.
As mentioned before the short selling ban has an significant influence on the cumulative abnormal return of the global systematically important financial institutions. During the short selling ban the cumulative abnormal return of the global systematically important financial companies was 92,01% lower, according to the estimation of the 𝑋𝑏𝑎𝑛,𝑡𝑋𝐺𝑆𝐼 coefficient. A possible explanation for this result can be that the global systematically important financial companies were more affected by the financial crisis, which lead to abnormal stock price declines.
Based on the regression results, the short selling ban has a significant impact on the standard deviation of the stock price return. During the short selling ban the volatility increased with 1,39%. This was in line with the hypothesis, based on the existing literature. The standard deviation is often used as a measurement of risk. Therefore these results imply that the risk during the short selling ban increased, measured as an increase of the standard deviation of the stock price return. If the standard deviation of the stock price return is used as measurement of market quality, the market quality worsens due to a short selling ban.
Finally, the results show us that the short selling ban has a significant influence on the volatility of the banned financial institutions. Only there is not a significant difference between the impact of the short selling ban on global systematically important financial institutions and the impact of the short selling ban on the other financial institutions. The expectation was that global systematically important financial institutions were significantly less affected by the short selling ban. So the risk of global systematically financial institutions as well as the risk of the other financial institutions increased due to the short selling ban.
24 5. Conclusion
This thesis investigated the impact of the UK 2008-2009 short selling ban on the stock prices and the volatility of the banned financial companies. In order to determine the impact of the short selling ban on the stock prices, the cumulative abnormal returns of the banned institutions are measured during the pre ban period, the ban period and the post ban period. Secondly, the volatility will be measured as the standard deviation of the daily stock price return for the pre ban period, the ban period and the post ban period. This thesis also focus on the difference between the short selling ban impact on global systematically important financial institutions and the impact on the other banned financial institutions.
Based on the existing literature the hypotheses were formed. The first hypothesis states that during the short selling ban, banned stocks have higher cumulative abnormal returns than when short selling is allowed. The short selling ban has a lower impact on the cumulative abnormal returns of the global systematically important financial institutions than the impact on the other banned institutions. According the second hypothesis, the standard deviation of the stock price return will increase due to the short selling ban. This effect will be lower for global systematically important financial institutions than for the other banned institutions.
To test the hypotheses a regression of the cumulative abnormal return and the standard deviation of the stock price return were performed. The results show that the cumulative abnormal returns increased during the short selling ban, but the increase was not significant. The results also show us that the cumulative abnormal returns of the global systematically important financial institutions are negatively affected by the short selling ban. Therefore the first hypothesis is rejected. The second regression provide evidence that the standard deviation of the stock price return will increase due to the UK short selling restrictions. The increase of the volatility was significant at a 5% level. Therefore the hypothesis, which state that the volatility will increase due to a short selling ban, is accepted. However, the hypothesis, which state that the short selling ban effect on the volatility of global systematically important financial institutions is lower than the effect on the volatility of the other banned institutions, is rejected. According the results, the short selling ban effect is lower for global systematically important financial institutions, but it was not significant lower.
There are things to keep in mind when drawing a conclusion in this thesis. First of all, is the size of the sample. This study only observe the impact of the short selling ban on 22 financial companies. Also there are only 3 observations of each company, namely before, during and after the short selling ban. Therefore the sample size is limited to draw statistical conclusion about the impact of the short selling ban. Secondly, the most recent list of Global Systematically Important Banks or Insurers is used in this thesis. In 2008 or 2009 the Financial Stability Board did not published a list of Global Systematically Important Banks or Insurers. Therefore the results can be biased, because in 2008 and 2009 the list of
25 Global Systematically Important Banks or Insurers can differ from the one that is used in this thesis. Further research could take look at the impact of UK short selling ban on the price discovery process. The results of this thesis suggest that the short selling ban has not a significant influence on the cumulative abnormal return, but Frino, Lecce and Lepone found evidence that the 10 day cumulative abnormal return after the implementation of the short selling ban significantly increased. This could indicate that the short selling ban worsens the price discovery process. Also further research can take a look at the difference of the short selling ban impact between different countries. Maybe the financial institutions of a particular country are more affected by a short selling ban than others.
26 References
Bai, Y., & Chang, E. C., and J. Wang. 2006. Asset prices under short-sale constraints. Working paper. Retrieved from the University of Hong Kong and MIT.
Beber, A., & Pagano, M. (2013). Short-Selling Bans around the World: Evidence from the 2007-09 Crisis. Journal of Finance, 68(1), 343-381.
Bodie, Z., & Kane, A., and Marcus, A. (2014). Investments (tenth edition). New York: McGraw-Hill education.
Boehmer, E., & Jones, C. M., and Zhang, X. (2013). Shackling short sellers: the 2008 shorting ban.
Review of Financial Studies, 26, 1363–1400.
Boehmer, E., & Wu, J. (2013). Short selling and the price discovery process. Review of Financial
Studies, 26, 287–322.
Boulton, J., & Braga-Alves, M. V. (2010). The skinny on the 2008 naked short-sale restrictions.
Journal of Financial Markets, 13, 397–421.
Brunnermeier, M.K., & Oehmke, M. (2014). Predatory short selling. Review of Finance, 18 (6), 2153-2195.
Charoenrook, A., & Daouk, H. (2005). The world price of Short Selling. Working paper. Retrieved from the Vanderbilt University and Cornell University.
Diamond, D.W., & Verrecchia, R.E. (1987). Constraints on short-selling and asset price adjustment to private information. Journal of Financial Economics, 18, 277-311.
Frino, A., & Lecce, S., and Lepone, A. (2011). Short-sales constraints and market quality: Evidence from the 2008 short-sales bans. International Review of Financial Analysis, 20, 225–236.
Helmes, U., & Henker, J., and 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.
Kraus, A., & Rubin, A. (2003). The Effect of Index Option Initiation on Volatility in the Presence of Heterogenous Beliefs and Short Sale Constraint. International Review of Finance, 4, 171-188. Levy, M., & Ritov, Y. (2001). Portfolio Optimization with Many Assets: The Importance of Short-Selling. Working paper. Retrieved from the eScholarship University of California.
27 Marsh, I. W., & Payne, R. (2012). Banning short sales and market quality: The UK’s experience.
Journal of Banking & Finance, 36, 1975-1986.
Miller, E. (1977). Risk, Uncertainty, and Divergence of Opinion. Journal of Finance, 32 (4), 1151-1168.
Saffi, P. A. C., & Sigurdsson, K. (2011). Price efficiency and short selling. The Review of Financial
Studies, 24, 821–857.
Staley, K.F. (1997). The art of short selling. Salt Lake city: John Wiley & Sons, Inc.
Woolridge, J.R., & Dickinson, A. (1994). Short Selling and Common Stock Prices. Financial Analyst
