1 All views expressed in the paper are views of the author and do not represent in any way the views of Cornerstone Research or Stanford Law School.
Informational Leakages: Evidence from short selling around
securities class action lawsuits
Thesis by:
Hans van Westrienen (6114350)
Supervised by:
Dr. P.J.P.M. Versijp
ABSTRACT
I examine whether short sellers are, in general, able to predict the filing of a securities class action lawsuit and in which month prior to, or after the filing of a securities class action lawsuit, the abnormal short interest is at its highest level. The abnormal short interest increases in the months prior to the filing of a securities class action lawsuit and reaches its highest point in the 5th month after the filing of a securities class action lawsuit. An upcoming filing of a securities class action lawsuit leads, in general, to an 2,1% increase in the abnormal short interest. Thereby, if a company is in the top 5% based on the abnormal short interest, the possibility of an upcoming securities class action lawsuit is almost doubled. These results indicate that short interest in general predicts a security class action lawsuit and facilitates the price discovery process around filings of a securities class action lawsuit.
2 There is a tremendous interest in short selling activity by the public. Several prominent financial papers wrote about short selling activity (Lamar, 2014). Critics of short selling argue that short sellers can make profits by spreading false information in an attempt to drive the share price down. Proponents of short selling counter the critics and argue that short selling increases the market efficiency and that it facilitates the price discovery process (Karpoff & Lou, 2010). Short sellers are professional, mostly institutional investors who sell borrowed stocks. They borrow stocks from an (institutional) investor and sell the stocks at the open market. After the decline in price, the short seller repurchase the stock (at the lower market price) and returns the stock to the original owner. Short selling is in essence a trading strategy that seeks to capitalize on a decline in the price of a security. A potential decline in the share price is the revelation of corporate fraud. Ferris & Pritchard (2001) proved the negative effect on the share price after the revelation of corporate fraud. This suggests that short selling and corporate fraud can play a role in market efficiency.
Corporate fraud, deliberate dishonesty to deceive the public, is a serious threat to investors. A lot of legislation is designed to prevent corporate fraud. A filing of a securities class action lawsuit is a filing of a claim to recover damages sustained as a result of corporate fraud. Prior to the adoption of the Private Securities Litigation Reform Act of 1995 – hereafter PSLRA – a lot of frivolous lawsuits were filed after a substantial drop in the share price (Johnson e.a., 2007). The PSLRA was designed to limit those frivolous lawsuits. After the adoption of the PSLRA, plaintiffs were required to bring forth the fraudulent acts of the accused company. A securities class action lawsuit against a company after the introduction of the PSLRA, is an allegation that the company violated federal or state security laws. A filed lawsuit leads to a negative stock market reaction of the subjected company (Ferris & Pritchard, 2001).
The negative stock market reaction after the revelation of a securities class action lawsuit is a potential profit for short sellers. As been discussed by Karpoff & Lou (2010), short sellers are shorting overvalued stock to expose the share price to the negative information known to the short seller. Chakrabarty and Shkilko (2013) proved that financial markets show informational leakages around insider trading. If an information leakage occurs prior to a securities class action lawsuit, short sellers might earn profits on the decline in share price. This research examines if those informational leakages are applicable to the announcement of a securities class action lawsuits. Hence the research question: “Does short selling increases prior to a securities class action lawsuit?”
This research contributes to the existing literature in several ways. First of all the research has significant insights in the ability of short sellers to detect and reveal securities class action lawsuits. Secondly, the study give useful insights in the informational leakages around securities class action lawsuits. Those insights are helpful for investors to determine their investment decisions and for
3 regulators to reconsider short sale (disclosure) policies to increase the market transparency. The research also addresses the efficient market hypothesis through a short selling perspective.
The results are clear. The potential profits of the filing of a securities class action lawsuit makes short sellers increase their short interest in lawsuit-companies. The average abnormal short interest in the month of the filing of the lawsuit is 2,1%. If a securities class action lawsuit is filed against a company, the possibility that this company is in the top 5% of the abnormal short interest measure is almost twice the possibility of when there is no securities class action lawsuit filed against the company. In other words, the possibility of an upcoming lawsuit is almost two times higher if the company is in the top 5% of the abnormal short interest measure. This provides the evidence that short selling leads to more price efficiency in the case of a securities class action lawsuit.
This paper is organized as follows. In Section I, the existing academic literature is reviewed. In Section II, the data and the short interest measures are explained. Section III explains the methodology of the research. Section IV examines the results. Section V is the conclusion.
I. Literature Review
Short Selling
There is a tremendous interest in short selling. As been discussed by Boehmer, Jones & Zhang (2008), an short seller is involved in at least 12,9% of the trades on the NYSE. Proponents and critics provide several arguments in favor to or against short selling. The proponents of short selling argue that short selling increases the market efficiency and the price discovery process (Chang, Cheng & Yu, 2007). Critics of short selling, in contrast, argue that short sellers are using manipulative and predatory trading strategies (Goldstein & Guembel, 2008). A lot of research examined short selling. The ability of short sellers to earn abnormal returns is widely examined in the academic literature.
Most articles find that short sellers are able to earn abnormal returns (Diamond and Verrecchia, 1987). The ability of short sellers to earn abnormal returns is mostly credited to the short selling of overvalued stocks (Diether e.a., 2009 and Boehmer & Wu, 2008). Christophe e.a. (2004) stated that short selling increases in the days prior to an earnings announcement. This suggests that short sellers have an informational advantage with respect to other investors and that they trade on their informational advantage prior to an earnings announcement. However, they stated that the pre-announcement short-selling is not a perfect indicator.
In contrary to the articles above, there are also some articles that refutes the ability of short sellers to earn abnormal returns (Woolridge & Dickinson, 1994). Henry & Koski (2010) find evidence
4 of short selling prior to seasoned equity offerings and they state that there is no increase in short selling around seasoned equity offerings. Even further, they stated that they documented a situation where short selling decreases the price efficiency. It is not perfectly clear whether in general, short sellers are able to earn abnormal returns.
Proponents of short selling are stating that short selling increases the price discovery process. Boehmer & Wu (2008) empirically demonstrated that short selling helps to keep the share prices in line with fundamentals. This is in line with several academic papers who also demonstrated that short selling leads to an increase in the price efficiency (for instance, Diamond & Verrecchia ,1987). Boehmer & Wu (2008) also find that the price distortion around large price changes is limited due to short selling. In contrast with the arguments of critics of short selling, they did not find any evidence that short sellers use manipulative trading strategies.
In contrast to Boehmer & Wu (2008), critics are stating that short sellers use manipulative trading strategies to increase their profits. An example of a manipulative trading strategy is to short sell the stock of a company and subsequently spread rumors about that company which will lead to a decline of the share price of the company. Some empirical evidence examined the manipulative trading strategies of short sellers. According to Safieddine & Wilhelm (1996), short sellers manipulate the price around seasoned equity offerings.
Several academic papers distinguish uninformed and informed traders (Diamond & Verrecchia, 1987; Boehmer & Wu, 2008). Uninformed traders (like market makers) are traders that trade on public information solely. Informed traders (like hedge funds) are traders that trade on private information. In my paper, no distinction will be made between informed and uninformed traders. The distinction will not have any effect on the results presented in my paper. I will examine whether short selling in general predict the filling of a securities class action lawsuit. Whether the short sellers are informed or uninformed traders does not give extra insights. In general the expectation is that the short sellers are mostly informed traders that predict the filing of a securities class action lawsuit or that short sellers are traders that try to diversify their portfolio by short selling stocks that have a positive correlation. If a portfolio of stocks have short and long positions in stocks with a positive correlation, the positive correlation will be reversed and turn into a negative correlation. This negative correlation leads to diversification and to a lower overall risk.
Corporate Fraud
Although corporate fraud is still not eradicated in the financial markets (Cornerstone Research, 2014), the effect of corporate fraud on shareholder wealth has not been examined widely in the existing literature. The difficulty to study the effect of corporate fraud is the absence of reliable data. Multiple datasets are available as a proxy for corporate fraud, but all these databases are
5 suffering from several biases (Karpoff e.a. , 2013). Still, there are several articles that examine the effect of corporate fraud on shareholder wealth. All the articles find a negative share price reaction to the revelation corporate fraud (Gande & Lewis, 2007 and Davidson e.a., 1994).
In a securities class action lawsuit, a company is sued to recover the losses of investors that occurred due to inflated share prices which are caused by the alleged fraud. The alleged fraud consists in the typical cases of misrepresented operations, financial performances or future prospects. Less frequently, investors are suing a company because they sold their shares for too little due to the price distortion (Ferris & Pritchard, 2001). In most cases, the company or their directors did not gain from the fraud. Nevertheless the directors as well as the company are sued to recover investors’ losses due to the (deliberate) misrepresentation. The losses are in most cases a substantial percentage of the market capitalization of the alleged company. Therefore, securities class action lawsuits have a substantial influence on the share price of a company.
In 1996 the Private Securities Litigation Reform Act of 1995 – hereafter PSLRA – was adopted. Prior to the adoption of the PSLRA a lot of frivolous lawsuits were filed after a substantial drop in the share price in the hope to find a sustainable claim . Several lawsuits were filed against companies ‘citing a laundry list of cookie-cutter complaints, typically based on a company’s prior optimistic statements that had not come to fruition’ (Johnson e.a., 2007). It is difficult to distinguish fraud and mere business. Plaintiffs’ lawyers have to rely on the limited publicly available information to decide whether to sue or not to sue. To discourage weak cases, the PSLRA, with a series of procedural barriers to file weak lawsuits, was adopted in 1996. Johnson e.a. (2007) proved that the introduction of the PSLRA led to a closer relation between factors related to corporate fraud and securities class action lawsuits.
Corporate fraud is a (deliberate) dishonesty to deceive the public. In case of a securities class action lawsuit, a lawsuit is filed to recover damages sustained as a result of the corporate fraud. As commented by Karpoff et all (2013), the term “fraud” has a colloquial and a technical meaning. By using securities class action lawsuits as a proxy for corporate fraud, it implies that the whole sample consists of illegal activities. Sorting fraud from mere business is difficult. Therefore, it is possible that in the data, some lawsuits are filed to a company that did not commit fraud. It is sometimes rather mere (but bad) business. It is still possible that investors suffer from the financial misrepresentation and that they file a securities class action lawsuit to recover their damages. Therefore, the term ‘fraud’ consists of more than only illegal activities. Therefore, in my paper, corporate fraud is used in a more colloquial way and also represents the filing of lawsuits that are dismissed by the court.
6 In an efficient market, all the information is publicly available and are reflected in the share prices. After the adoption of the Sarbanes-Oxley act of 2002, an insider trade (a stock transaction done by an executive of the company) has to be submitted to the SEC within two business days following the transaction. All the insider trades are made publicly available by the SEC. Chakrabarty & Shkilko (2013) found evidence that insider stock sales are often accompanied by abnormal short selling in the same stock. This abnormal short selling is prior to the revelation of the insider trade(s) to the public and sometimes even before the insider trade is completed. They attribute this abnormal increase to informational leakages around the insider trades.
Efficient Market Hypothesis
It has been argued for a long time that the financial markets are efficient (Fama, 1970). In an efficient financial market, ‘security prices fully reflect available information in a rapid and unbiased fashion and thus provide unbiased estimates of the true underlying value’ (Basu, 1977). If the financial markets are efficient, short sellers should be unable to predict firms against which a securities class action lawsuit will be filed. This research does not examine whether the efficient market hypothesis holds, but it addresses the efficient market hypothesis through a short interest perspective.
Related studies
The key articles most related to mine are of Karpoff & Lou (2010) and of Chakrabarty & Shkilko (2013). Karpoff and Lou (2010) examined if short sellers detect firms that misrepresent their financial statements. They discovered that short selling leads to a faster time-to-discovery and that the shares are less overpriced if firms are misrepresenting their financial statements due to short selling. Chakrabarty & Shkilko (2013) found that there are informational leakages in financial markets. Their evidence is based on insider trading. They document considerable increases in short selling on days when company insiders sell their stock. This suggests informational leakages prior to insider trading.
The research of Karpoff & Lou (2010) and of Chakrabarty & Shkilko (2013) are the key academic papers to my study. The informational leakages of Chakrabarty & Shkilko is examined in my study as well. The difference is that my study focusses on corporate fraud instead of insider trading. The method used by Karpoff and Lou (2010) is used in my study as well. They determine in their research the abnormal short selling in stocks that misrepresent their financial statements.
This study will examine the abnormal short interest in firms against which a securities class action lawsuit is filed. This research contributes to the existing literature in several ways. First of all the research has significant insights in the ability of short sellers to detect and reveal securities class action lawsuits. Secondly, the study gives useful insights in the informational leakages around
7 securities class action lawsuits. Those insights are helpful for investors to determine their investment decisions and for regulators to reconsider short sale (disclosure) policies to increase the market transparency. The research also addresses the efficient market hypothesis through a short selling perspective. Lastly, this study addresses the discussion about short sale disclosure policies and price efficiency around securities class action lawsuits.
II. Data
In this analysis use is made of five different datasets; the Securities and Exchange Commission’s Regulation SHO, the Center of Research on Security Prices, the CompuStat database, the Options Clearing Corporation and the Securities Class Action Clearinghouse (SCAC) database of securities class action lawsuits.
Short interest
The Securities and Exchange Commission’s Regulation SHO is used to collect monthly data of short positions. This data is publicly available and publishes the short positions in stocks every day. The stocks of several firms are not exchanged on (a regulatory) stock exchange and therefore there is no short interest data. All firms without consistent short interest data are excluded from the research.
Securities class action lawsuits
The Securities Class Action Clearinghouse (SCAC) database is a publicly accessible database of securities class action lawsuits since 1996. The data is provided by Cornerstone Research and Stanford Law School. The data will be obtained for the period January 1996 – December 2012. The time period starts at 1996 because the SCAC is only available since 1996. The period ends in December 2012 to be sure that all the securities class action lawsuits are proceeded, settled or dropped after the filing. The data consists of 2,159 class action lawsuits between 1996 and 2012, dated by the filing date of the lawsuit. The filing date of the securities class action lawsuit made it possible to focus on short selling around this date.
The SCAC and the SHO databases are combined with CRSP and CompuStat databases to control for firm specific factors. The Options Clearing Corporation is used to determine which company has listed stock options.
In table I, an overview is given of all SCAC lawsuits filed in every year since 1996. The lawsuits used in this paper are in panel B. The securities class action lawsuits which are used in this paper differs from the filed lawsuits. There are several reasons for the difference in lawsuits between the number of cases which are filed and the number of cases after the restrictions. First of all, the filed
8 lawsuits against companies without short interest data are not included. If a company against which a securities class action lawsuit is filed does not have data about its short interest, it is impossible to determine whether the short interest increases prior to the filing of the lawsuit. Therefore, companies with missing data about short interest are omitted from the study. Secondly, for some of the companies there is missing data for (one of the) control variables to determine the expected short interest. If there is missing data, the company is not included into the dataset. After all the restrictions, 706 lawsuits remains and are used to determine the abnormal short interest. This is approximately 34% of the filed lawsuits. The percentages of used lawsuits in the study compared to the filed lawsuits per year are in panel C. It is clear in the table that the majority of the used lawsuits are of the lawsuits filed in the last decade.
Variables Mean Median Standard Error Minimum value Maximum value
Panel A
Short Interest 0,038 0,019 0,055 0 0,986
Momentum 0,014 0,005 0,537 -0,972 308,63
Trading volume 211.495 37.708 1.204.442 1 201.000.000 Market Capitalization 3.94e+09 5.13e+08 1.69e+10 63.217 6.27e+11 Panel B
Short Interest 0,052 0,028 0,067 0 0,918
Momentum 0,014 0,006 0,533 -0,902 127,45
Trading volume 598.183 139.929 2.950.824 67 201.000.000 Market Capitalization 1.15e+10 1.45e+09 3.55e+10 1.759.500 6.27e+11
Table II
Describtive statistics of selected variables.
This table gives the descrivtive statistics of several variables which are used to determine the abnormal short interest. The short interest is the short position in a stock divided by the shares outstanding in that stock. Momentum is the relative difference to the share price in the prior month. Trading volume is the amount of shares that is traded in each month. Market Capitalization is the market capitalization of the company in the specific month. The mean, median, standard error, minimum and maximu m values are given for each variable. The observarion which are taken into account are after the restrictions. For instance, if no short interest data is available, the observarions are dropped from the sample. Panel A consists of all the observarions after the general restrictions,
9 All views expressed in the paper are views of the author and do not represent in any way the views of Cornerstone Research or Stanford Law School.
Year 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Total Panel A Cases 79 122 158 135 137 179 160 142 151 122 97 125 146 122 125 126 105 2,159 Panel B Restricted 4 8 9 21 20 13 36 39 67 60 54 70 71 56 66 57 55 706 Panel C Percentage 5% 7% 6% 16% 15% 7% 23% 27% 44% 49% 56% 56% 49% 46% 53% 45% 52% 34%
Table I
Describtion of securities class action lawsuits sample
This table describes the number of securities class actions lawsuit cases filed in every year since 1996 through 2012. Panel A consists of all the la wsuits filed per year. Panel B consists of the securities class action lawsuit cases after restrictions. Some restrictions are taken into acount to control for missing data. If a company has, for instance, no data about the short interest in that firm, the cases are omitted from the study. In Panel B, all the companies with missing data for short interest, stoc k price, total assets, shares outstanding, trading volume or ticker are
10 All views expressed in the paper are views of the author and do not represent in any way the views of Cornerstone Research or Stanford Law School.
Table II consists of the descriptive statistics of the most important variables which are used. The observations are restricted to the same restrictions as the short interest above. Panel A consists of all the observations after the restrictions. Panel B consists of only the observations after the restrictions when a securities class action lawsuit will be filed. Table II shows that in general the market capitalization is higher if a securities class action lawsuit is filed. This also holds for the trading volume and the short interest. An explanation for these observations is that the securities class action lawsuit are in most cases filed against larger companies. More surprisingly is the fact that the statistics suggests that the monthly return in Panel B is almost equal to the monthly return in Panel A. This can be explained by the fact that Panel B consists of the companies in the months prior to the filing of a securities class action lawsuit. These results suggests that in general the monthly return of companies prior the filing of a securities class action lawsuit is equal to the monthly return of all companies in the sample.
III. Methodology
According to the literature there is one main hypothesis. The hypothesis of my paper is that short selling increases prior to the filing of a securities class action lawsuit. This hypothesis is based on Karpoff & Lou (2010) and on Gande & Lewis (2007). Karpoff & Lou (2010) found that short sellers detect companies that misrepresent their financial statements. This hypothesis is similar to their findings. Gande & Lewis (2007) found that there is a negative share price reaction on shareholder initiated class action lawsuits. The negative share price reaction is a potential profit for short sellers. This potential profit leads to an increase in short selling prior to the filing of a securities class action lawsuit. This hypothesis has two possible explanations. First of all, if the Efficient Market Hypothesis holds (Fama, 1970), the effects of upcoming filings of securities class action lawsuits are already reflected in the price. It would be, in general, impossible for short sellers to make a profit by predicting filings of securities class action lawsuits. It is possible that the Efficient Market Hypothesis does not hold in this case. There is a possibility that short sellers are better predictors of the filings of securities class action lawsuits than the market and that short sellers are making abnormal profits with their predictions. This would denote a breach of the Efficient Market Hypothesis. According to the Efficient Market Hypothesis it is impossible that abnormal returns are earned with use of public information. If they do not have any private information, abnormal returns earned by short sellers denote a breach of the Efficient Market Hypothesis. This view is labeled the prediction hypothesis (after Christophe e.a., 2008). On the other hand it is possible that there are informational leakages to short sellers about the filing of securities class action lawsuits. This view is labeled the tipping hypothesis. There is one main paper in favor of the tipping hypothesis. According to Chakrabarty & Shkilko (2013) there is an informational leakage regarding insider trades. They proved that short
11 selling increases prior to insider trades. This suggest informational leakages and rules out that informational leakages do not exist in the financial markets. Therefore, I expect that informational leakages also occurs prior to the filing of securities class action lawsuits.
The hypothesis will be tested in several ways. First of all, a standard OLS regression of the short interest in a firm and a dummy for a securities class action lawsuit will be used.
(1) In the regression, the short interest per month in firm, SIi,t, is regressed on a dummy variable
SCAL, which is equal to one if a securities class action lawsuit is filed against a firm. The regression is controlled for several control variables and firm specific factors. Several articles examined the effect of those control variables to short selling (Dechow e.a., 2011, Einhorn, 2010 & D’Avolio, 2002) and those control variables are also used by Karpoff & Lou (2010) to determine the abnormal short interest. The control variables are market capitalization, book-to-market ratio, past performance, industry and share turnover. Those control variables are somehow related to the short interest in a firm and therefore the equation controls for those factors. For instance, the past performance is related with short interest. If the past performance of a company is negative, it is more likely that short sellers shorted that company to make a profit. The opposite holds for companies with a positive past performance. To control for those factors, the control variables are included in the regression. For the control variables, all the firms are assigned to a portfolio which is low, medium or high (except for industry and share turnover). Those portfolios are dummy variables which are equal to one if a firm is assigned in that portfolio. The coefficient of interest in this regression is β1. My expectation is that β1
is significantly positive. This would prove that the short interest in a firm is higher if a securities class action lawsuit is filed against a firm.
In addition to the raw short interest regression, the abnormal short interest is examined as well. For every stock, the abnormal short interest is determined based on a several control variables.
[ ] (2)
In this regression, SI i,t represents the short interest in stock i in month t. E [SIi,t] represents the
expected short interest in stock i in month t. The expected short interest is based upon the same control
variables as in equation 1.
[ ] (3) The abnormal short interest will be measured during a 40-month period surrounding the public revelation of the fraud. This time frame is adopted from Karpoff & Lou (2010). They found that the abnormal short interest in firms that misrepresent their financial statements is statistically
12 significant between the 17th month prior to the revelation till the 5th month after the revelation. In this study the revelation can differ from the filing of the securities class action lawsuit. This is why the time frame is 40-months instead of 24-months.
With use of the abnormal short interest, several measures are used to examine the effect of a securities class action lawsuit on the (abnormal) short interest in a firm. Equations (4), (5) and (6) show those regressions.
(4)
(5)
(6) By determining the abnormal short interest in stocks that suffer from a securities class action lawsuit, the research will examine whether the short interest is higher in stocks that suffer from such lawsuits. If the hypotheses holds and the short interest is higher, short sellers know about the upcoming securities class action lawsuit. This suggests that there are informational leakages about the securities class action lawsuits.
Determining the abnormal short interest per firm does not confirm the hypothesis whether securities class action lawsuits are in general predicted by the abnormal short interest. To determine if high levels of short interest are related to the filing of a securities class action lawsuit all firms are assigned to a portfolio based on their level of abnormal short interest. A firm is classified as a high short interest firm if the ABSI is in the top 5%. If the ABSI is in the bottom 95%, the firm is classified as a low short interest firm. The presence of firms in the high short interest group against which a securities class action lawsuit is filed is determined and compared with the presence of firms against which a securities class action lawsuit is filed in the low short interest group. If the short interest is higher if a securities class action lawsuit is filed, the percentage of firms against which a securities class action lawsuit is filed should be significant higher in the high short interest group. This would in general predict that high levels of short interest are related with the filing of a securities class action lawsuit. Therefore, the hypothesis that short interest in general predicts the filing of a securities class action lawsuit holds. A chi squared test will be used to determine if the hypothesis that short interest predicts securities class action lawsuits will be accepted.
Endogeneity
In the different specifications there are two potential problems. First of all, there is a possibility of omitted variables. It is not perfectly clear which variables have an influence on the short interest. Therefore, the estimations of the ABSI may be imperfect and suffers from omitted variable bias.
13 Secondly, there is a potential endogeneity problem. In equation (3), the expected short interest is determined based on several variables including market-to-book ratio and market capitalization. Both variables are somehow based on the share price of the company. As demonstrated by Boehmer & Wu (2008), short selling is related with the share price of a company. In this case, the specifications suffer from simultaneous causality. It is possible that the market-to-book ratio is lower due to the short selling rather than that the short interest is driven by a low market-to-book ratio.
In equations (1) – (6), the problem is that the abnormal short interest is measured with use of the same set of variables which are used as control variables. Equation (6) is basically the same as equation (1) without the control variables. To address this endogeneity problem, an instrument is used to determine the abnormal short interest in a different way.
(7)
Following Karpoff and Lou (2010), Options is a dummy variable set equal to one if the stock has listed stock options. The relevance of tradable options as an instrument is demonstrated by Diether e.a. (2009), by finding that short selling volume has a positive relation with the availability of options through the cost of hedging short positions. If a stock has listed options, it is less costly to hedge a short position. Therefore, the abnormal short interest in a firm is in general higher if a stock has listed stock options. However, the relation between the two is relatively low. The correlation is approximately 23% which is relatively low to use it as an instrument. Even though the correlation is relatively low, it is still a valid instrument. The effect of listed stock options on the abnormal short interest is significant and positive. Therefore, the abnormal short interest is determined using the fitted values of equation (7).
̂ ̂ ̂ (8)
The fitted values of equation (7) are in equation (8). With use of equation (8), the ABSIi,t is
determined. This ABSIi,t is used to determine whether a securities class action lawsuit leads to higher
short interest. This specification is specified in equation (9).
̂ (9)
In equation (9), the ABSIi,t is based on an intercept and a dummy variable that equals 1 if a
securities class action lawsuit is filed. The ABSIi,t is the fitted value of equation (8). The coefficient of
interest in this case is α1 which determines the effect of the securities class action lawsuit on the
abnormal short interest. There is an effect of a securities class action lawsuit if α1 is significantly
positive. The endogeneity problem in this equation is an omitted variables problem. Without any control variables, it is possible that the effect of a securities class action lawsuit is biased. To determine an unbiased estimator of α1 some firm specific factors have to be added. In equation 6
14 several control variables are added which are in line with previous literature and have a significant effect on the abnormal short interest. In addition to equation (9), a chi-squared test will be done to examine the effect of an securities class action lawsuit to the short interest in a firm. Based on the fitted values of equation (8), all firms are assigned to a portfolio based on their level of abnormal short interest. A firm is classified as a high short interest firm if the ABSI is in the top 5%. If the ABSI is in the bottom 95%, the firm is classified as a low short interest firm. A Chi-squared test will be used to determine if the null hypothesis that short interest predicts securities class action lawsuits will be accepted.
The instrument which is used to omit the possible endogeneity problem is listed options. This is a dummy variable and the only explanatory variable which is used to determine the abnormal short interest. Because it is the only explanatory variable there is a possibility that the listed options are a weak instrument. Therefore another instrument is used to determine the abnormal short interest in a different way. The instruments which are used are the control variables used to determine the expected short interest. In equation 1 the expected short interest is determined based on several variables which affect the short interest. The first stage is the equal to equation 1, the second stage of the two stage least squares regression is in equation 12. The difference with equation 4 is that the abnormal short interest is based on the predicted value of the regression of equation 1.
̂ ̂ (10) ̂ (11)
The advantage of the two stage least squares regression is that the endogeneity problem is resolved. The endogeneity problem exists due to the fact that the short interest might affect the share price. This share price is used to determine several of the control variables (market capitalization and Book-to-Market value for instance). By taking only the predicted values in the first stage of the two stage least squares regression, the endogeneity problem resolves itself. Short interest is a good instrument because it is highly correlated (93%) with the abnormal short interest.
IV. Results
Short sellers are making profits by selling borrowed stocks. By selling those stocks they try to benefit from a decline in the share price after their short sale. After the decline in the share price, the short sellers buy the shares back at a lower price and return the shares to the original owner. It is key for short sellers that the share price declines, otherwise their trade will not be profitable.
15 In this paper, I demonstrate that short selling increases prior to a securities class action lawsuit. The reason that short selling increase prior to a securities class action lawsuit is the fact that short sellers can make money due to the decline in share prices caused by the filing of a securities class action lawsuit. To determine whether short selling increases prior to and around a securities class action
lawsuit, it is key to determine whether the condition are in favor of short sellers, i.e. does the share price declines around the filing of a securities class action lawsuit. Figure I shows the monthly return in a 40-month event window around the filing of a securities class action lawsuit.
As is demonstrated in figure I, the share price has an average decline of 10% in the month of the filing of the securities class action lawsuit. This gives a favorable environment for short sellers to short sale companies prior to the filing of a securities class action lawsuit. A minor point of discussion is the fact that it is the raw return and not the abnormal return. In my view, short sellers are not specifically interested in the abnormal return. Most short sellers are making a profit by shorting stocks that have a negative return. Whether this negative return is abnormal or just the raw return does not matter. Therefore, the raw return is used instead of the abnormal return. Thereby, short sellers do not predict the abnormal return after the filing of a securities class action lawsuit. Short sellers are interested in the negative stock market reaction to the filing of the lawsuit. The raw return has to be negative to profit from the trade. They do not care about the abnormal return.
Figure I gave a broad view of the average return in the months around a securities class action lawsuit. As is demonstrated, those returns are presumably negative and that leads to potential profits for short sellers. The question is, are short sellers able to act on those potential profits? To give a broad view of
-15,00% -10,00% -5,00% 0,00% 5,00% 10,00% -20 -15 -10 -5 0 5 10 15 20 Re tu rn
Months relative to the securities class action lawsuit filing
16 0,00% 0,50% 1,00% 1,50% 2,00% 2,50% 3,00% -20 -15 -10 -5 0 5 10 15 20 A b n o rm a l s h o rt in te re st
Months relative to the securities class action lawsuit filing
Figure II: Abnormal short interest in a 40-month event window around the filing of a class action lawsuit
the abnormal short interest around the filing of a securities class action lawsuit, the abnormal short interest is determined in the same 40-month event window as the returns are.
Figure II shows that the abnormal short interest increases steadily till it reaches its highest point (2,5%) in the sixth month after the filing of the lawsuit. After this month the abnormal short interest decreases over time. The abnormal short interest is differs significantly from zero from 19 months prior to the filing of the lawsuit till the 19th month after the filing of the lawsuit. Only the 18th month after the filing of the lawsuit does not significantly differ from zero within this timeframe (at a 5% level). The short interest is not fully explained by the companies trading characteristics like the trading volume and market capitalization around securities class action lawsuits. It is plausible that short selling is in general predicted by other characteristics around securities class action lawsuit filings. This suggests that there is abnormal short interest in the companies prior to and around securities class action lawsuits.
The steady increase in the short interest prior to the filing of a securities class action lawsuit was expected. The increase in the five months after the filing of the securities class action lawsuit was unexpected. It is possible that different short sellers have to be distinguished in this case. It is possible that after the filing of the securities class action lawsuit, new short sellers opened new short positions. Due to the collectivity of the used dataset, there is no distinguish between different short sellers and no evidence is found that supports this hypothesis. Another possibility is that it takes time to unwind the (sometimes) large short positions.
To support the hypothesis and the suggestions triggered by figure I & II that short selling increases prior to the filing of a securities class action lawsuit, several tests are conducted. The results of the tests are presented in three different ways. At first, the effect of a securities class action lawsuit
17 on the raw short interest is determined. Subsequently, the effect of a class action lawsuit on the abnormal short interest in determined and lastly the Chi-squared will test whether class action lawsuit firms are more frequently in the 5% highest abnormal short interest than non-class action lawsuit firms.
In table III, the effect of a securities class action lawsuit on the raw short interest is given. The coefficient of interest is in panel C. The effect of a securities class action lawsuit on the raw short interest is 1.2%. This means that a securities class action lawsuit leads to a higher short interest of 1.2%. The coefficient is significant at a 1% level. This provides evidence that a securities class action lawsuit affects the raw short interest in a positive way. In table III the securities class action lawsuit is a dummy variable that equals one if a securities class action lawsuit is, will be or has been filed against a company. Table III provides empirical evidence that a securities class action lawsuit affects the raw short interest, but it doesn’t provide any evidence whether short sellers are able to predict the filing of a securities class action lawsuit.
18
Panel A Panel B Panel C Panel D Panel E Panel F
Variable name SI (1) SI (2) SI (3) ABSI (1) ABSI (2) ABSI (3)
SCAL 0.024 0.020 0.012 0.029 0.024 0.012 (14.11)** (12.81)** (8.76)** (14.91)** (13.97)** (8.76)** MarketCapLow -0.023 -0.015 -0.019 -0.000 (85.70)** (56.67)** (64.14)** (0.00) MarketCapMed -0.005 -0.001 -0.006 -0.000 (22.00)** (7.06)** (27.85)** (0.00) MomentumLow 0.002 0.002 -0.001 -0.000 (14.53)** (16.40)** (6.44)** (0.00) MomentumMed -0.000 0.002 -0.004 -0.000 (2.40)* (15.40)** (24.80)** (0.00) BtMLow -0.013 -0.016 0.004 0.000 (50.32)** (62.50)** (14.20)** (0.00) BtMMed -0.009 -0.010 0.001 0.000 (46.63)** (51.76)** (6.49)** (0.00) VolumeLow -0.033 -0.066 (176.10)** (352.20)** VolumeMed -0.021 0.000 (146.38)** (0.00) Iaff 0.013 0.000 (1.62) (0.00) Imining 0.009 0.000 (1.80) (0.00) Icons 0.023 0.000 (3.78)** (0.00) Imanu 0.020 0.000 (4.32)** (0.00) Itcegs 0.008 0.000 (1.68) (0.00) Iwhtr 0.008 0.000 (1.61) (0.00) Iretr 0.030 0.000 (6.28)** (0.00) Ifir 0.003 0.000 (0.69) (0.00) Iser 0.020 0.000 (4.31)** (0.00) _cons 0.032 0.051 0.053 -0.027 -0.017 -0.000 (61.30)** (94.56)** (11.73)** (46.42)** (28.28)** (0.00) N 586,926 586,926 586,926 586,926 586,926 586,926 * p <0.05; ** p <0.01 Table III
Regressions of SI & ABSI
This table reports the coefficients for the regression in equation 1, 4, 5 & 6 (Panel A, B, C, D, E & F). SI is the short interest in a stock and is determined by the number of shares shorted in a company as a percentage of the total shares outstanding. ABSI is the abnormal short interest in a stock and is determined as in equation 2. SCAL is a dummy
variable that equals one if a securities class acction lawsuit will be filed against the company . MarketCap, Momentum, Book-to-Market and Volume are variables that consists of several portfolios. A firm is qlassified as a high
(i.e. VariablenameHigh = 1) if the firm is in the highest portfolio for that variable. The industry variables (starting with "I") are dummies for n-1 industries and equals 1 if the company acts in the industry. T -statistics test whether the
19 To determine whether short selling in general predicts the filing of a securities class action lawsuit, a chi-squared test is conducted to test if the presence of abnormal short interest is related with an upcoming filing of a securities class action lawsuit. Table IV present the results of the chi-squared test. The percentage of lawsuit-companies in the high abnormal short interest group is 9.52%, while the percentage of non-lawsuit companies in the high abnormal short interest group is only 4,68%. The chi-squared statistic is 1.700 and is significant at a probability of 0.000. This test provides empirical evidence that there are relatively more companies in the high abnormal short interest group against which a securities class action lawsuit will be filed. This gives the evidence in favor of the hypothesis and to conclude that in general, high abnormal short interest predicts an upcoming filing of a securities class action lawsuit.
ABSI High (>95%) ABSI Low (<95%) Total Lawsuit Frequency 3,443 32,721 36,164 Percent 0,59% 5,57% Row % 9,52% 90,48% Column % 11,77% 5,87% No Lawsuit Frequency 25,802 524,960 550,762 Percent 4,40% 89,44% Row % 4,68% 95,32% Column % 88,23% 94,13% Total 29,245 557,681 586,926
Chi-squared statistic 1.700 p-value 0.000
* Percentages m ight not add up to 100 due to roundings
Table IV
Chi-squared test
This table reports the Chi-squared test of the relation between securities class action lawsuits and ABSI. ABSI is the abnormal short interest in a stock and is determined by the actual short interest minus the expected short interest. If a company 's ABSI is in the highest 5%, the company is
classified as a "high ABSI" company. If a company's ABSI is in the lowest 95%, the company is classified as a "low ABSI" company. If a securities class action lawsuit is filed against a company in a company, all
the months prior to the filing of the lawsuit are classified as a "lawsuit" company. If no lawsuit is filed against a company or if a lawsuit is filed in
a prior month, the company will be classified as a "no lawsuit" company. The table consists of the frequency, the percentage of the total amount of observations, the percentage of the total amount of observations in its row and the percentage of total amount of observations in its column. The
Chi-squared statistic tests the null hypothesis that short interest predicts securtities class action lawsuits.
Pearson chi2(1) = 1.7e+03 Pr = 0.000 Total 557,681 29,245 586,926 1 32,721 3,443 36,164 0 524,960 25,802 550,762 L 0 1 Total PriorToSCA ABSIhigh 93.84 6.16 100.00 Total 550,762 36,164 586,926 88.23 11.77 100.00 1 25,802 3,443 29,245 94.13 5.87 100.00 0 524,960 32,721 557,681 ABSIhigh 0 1 Total PriorToSCAL
20
Panel A Panel B Panel C Panel D Panel E Panel F
Variable name ABSI (1) ABSI (2) ABSI (3) ABSI (1) ABSI (2) ABSI (3)
SCAL 0.017 0.014 0.006 0.005 0.002 0.001 (78.50)** (62.76)** (29.63)** (94.51)** (39.12)** (29.44)** MarketCapLow -0.022 -0.003 -0.016 -0.013 (123.23)** (14.03)** (387.52)** (289.18)** MarketCapMed 0.013 0.019 -0.005 -0.004 (71.82)** (113.24)** (122.82)** (105.79)** MomentumLow 0.003 0.003 0.000 0.000 (18.55)** (19.44)** (9.13)** (9.00)** MomentumMed -0.004 0.002 0.000 0.001 (25.36)** (11.56)** (8.63)** (19.21)** BtMLow -0.001 -0.004 (3.13)** (92.52)** BtMMed -0.006 0.001 (36.23)** (20.60)** VolumeLow -0.051 0.000 (268.68)** (6.88)** VolumeMed -0.034 0.000 (202.96)** (5.06)** Iaff 0.020 0.012 (11.11)** (27.52)** Imining 0.005 0.011 (3.51)** (32.93)** Icons 0.024 0.015 (16.26)** (41.09)** Imanu 0.015 0.009 (10.82)** (28.62)** Itcegs 0.010 0.010 (7.40)** (31.03)** Iwhtr 0.011 0.010 (7.78)** (30.18)** Iretr 0.028 0.010 (20.07)** (30.16)** Ifir 0.008 0.008 (6.17)** (25.21)** Iser 0.017 0.008 (12.27)** (25.87)** _cons -0.023 -0.019 -0.011 -0.021 -0.014 -0.023 (284.46)** (112.09)** (8.00)** (1,070.86)** (371.07)** (71.68)** R2 0.01 0.07 0.20 0.01 0.23 0.25 N 586,926 586,926 586,926 586,926 586,926 586,926 * p <0.05; ** p <0.01 Table V
Instrumental regressions of ABSI
This table reports the coefficients for the instrumental regressions in equation 9, 10 & 11 (Panel A, B, C, D, E & F).
In Panel A, B & C the instrument which is used to determine the abnormal short interest is the short interest. In
Panel D, E & F, listed options is used as an instrument to determine the abnormal short interest. SCAL is a dummy
variable that equals one if a securities class acction lawsuit will be filed against the company. MarketCap, Momentum, Book-to-Market and Volume are variables that consists of several portfolios. A firm is qlassified as a
high (i.e. VariablenameHigh = 1) if the firm is in the highest portfolio for that variable. The industry variables
(starting with "I") are dummies for n-1 industries and equals 1 if the company acts in the industry. N is the total
number of observations and is divided between months and companies. 6730 companies are used. The number of
21 As explained in section IV, the abnormal short interest is determined in two ways as given in equation 10 and 11. To rule out possible endogeneity problems, an instrument is used to determine abnormal short interest in an alternative way. This instrument is the availability of listed stock options. If a company has listed and tradable options, it is easier and less costly to hedge a short position in that company. The relevance of listed options as an instrument is demonstrated by Diether e.a. (2009). Karpoff and Lou (2010) also used listed options as an instrument to determine the abnormal short interest. However, the relevance of listed options as a valid instrument is doubtful due to the low correlation between the abnormal short interest and listed options (23%). Therefore, next to listed options as an instrument, the control variables are used as an instrument as well. The results are in table V.
Table V consists of the instrumental variable regressions of the abnormal short interest. The evidence in table V indicates that a filing of a securities class action has an effect on the abnormal short interest. All the coefficients are significant (p < 0.01) and are positive. The results in Panel D, E & F indicate that if the listed stock options of a company are used as an instrument, the coefficient of the filing of a securities class action lawsuit are relatively low. There is less than a half percent increase in the abnormal short interest if a securities class action lawsuit will be filed. The results in Panel A, B & C indicate that an upcoming filing of securities class action lawsuit leads to at least a half percent increase in the abnormal short interest. The results are influenced by the fact that all the data for the company prior to the filing of the securities class action lawsuit are taken into account. The abnormal short interest is significantly positive from the 18th month prior to the filing of the lawsuit. Companies are considered ‘lawsuit-companies’ if a securities class action lawsuit will be filed in the future. It is possible that the abnormal short interest is negative two or three years prior to the filing of the lawsuit. Those observations are still taken into account to determine the effect of the securities class action lawsuit.
The results provide evidence that short interest increases prior to the filing of a securities class action lawsuit. Boehmer & Wu (2008) already demonstrated the negative price effect of short selling. They provided evidence that short selling decreases the share price and therefore limited the price distortions around large price changes. This effect of short selling also exists around securities class action lawsuits. This suggests that short selling around securities class action lawsuits increases the price efficiency.
V. Conclusion
Short selling is a main topic in several discussions about the financial markets. Critics of short selling argue that short sellers are the one of the main causes of declined share prices. Proponents of short
22 selling argue that short selling leads to more price efficiency and that it facilitates the price discovery process. My thesis provides empirical evidence on one minor aspect of short selling; does short selling increase around the filing of a securities class action lawsuit and does short selling predict the filing of a securities class action lawsuit. The hypothesis is based on the fact that the filing of a securities class action lawsuit leads to a negative price effect. This negative price effect is a potential profit for short sellers. The average monthly return in the month of the filing of a securities class action lawsuit is a decline of more than 10%. The abnormal short interest in companies against which a securities class action lawsuit is filed increases steadily till it reaches its highest point (2,4%) in the sixth month after the filing of the lawsuit. The abnormal short interest is significant since the 19th month prior to the filing of the lawsuit till the 17th month after the filing of the lawsuit. I did not provide any evidence why the abnormal short interest remains positive after the filing of the lawsuit. Possible explanations are that it cost time to unwind the short positions, that there are different short sellers with different strategies so it is possible that new short positions are opened after the filing of the lawsuit and it is possible that short sellers anticipate on the negative price reaction on the outcome of the lawsuit.
These results suggests that short sellers are able to anticipate on the filing of a securities class action lawsuit. This suggestion is confirmed by the difference in frequency of the lawsuit- and non-lawsuit companies in the abnormal short interest portfolios. It is more likely that companies against which a securities class action lawsuit is filed are in the top 5% of the abnormal short interest measure. On average, 9,52% of the companies against which a securities class action lawsuit will be filed are in the top 5% of the abnormal short interest measure. This is almost two times the quantity which would be expected if there is nog relation between securities class action lawsuits and abnormal short interest.
These results indicate that short sellers are in general able to predict the filing of a securities class action lawsuit. The potential returns on the filing of such a lawsuit are such that the abnormal short interest increases (on average) to 2,1% of the shares outstanding. By short selling the companies against which a lawsuit will be filed, they decrease the share price which leads to a lower decline in the share price in the month of the filing of the lawsuit. This provides empirical evidence that in this case short selling leads to more price efficiency.
The results are robust to two instrumental variable regressions. If listed stock options and short interest are used as an instrument to determine the abnormal short interest, the possibility of omitted variable biases and endogeneity problems are resolved. The results are identical to the results without the instruments. The filing of a securities class action lawsuit still leads to higher abnormal short interest levels. However, the increase in the abnormal short interest levels is in general a bit less
23 than without the instruments; I found an average increase of more than a half percent if a securities class action lawsuit will be filed.
The results all point out that the filing of a securities class action lawsuits leads to more short selling and that, in general, upcoming securities class action lawsuit filings are predicted by short selling and the increase in the short interest. The average share price decline in the month of the filing of the securities class action lawsuit is more than 10%. Therefore, short sellers profit from the short positions in the companies against which a securities class action lawsuit will be filed. The increase in short interest leads to more price efficiency and therefore facilitates the price discovery process.
24
Bibliography
Basu, S. (1977). Investment performance of common stocks in relation to their price‐earnings ratios: A test of the efficient market hypothesis. The Journal of Finance, 32(3), 663-682.
Boehmer, E., Jones, C. M., & Zhang, X. (2008). Which shorts are informed?. The Journal of Finance, 63(2), 491-527.
Boehmer, E. and J. Wu. (2008). “Short Selling and the Informational Efficiency of Prices.” Working paper, University of Georgia.
Chakrabarty, B., & Shkilko, A. (2013). Information transfers and learning in financial markets: Evidence from short selling around insider sales. Journal of Banking & Finance, 37(5), 1560-1572.
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.
Christophe, S. E., Ferri, M. G., & Hsieh, J. (2010). Informed trading before analyst downgrades: Evidence from short sellers. Journal of Financial Economics, 95(1), 85-106.
Christophe, S., M. Ferri, and J. Angel. (2004) “Short selling Prior to Earnings Announcements. ”Journal of Finance 59, 1845-1875.
Cornerstone Research. (2014). Securities class action filings 2013: a year in review.
Davidson, W., D. Worrell, and C. Lee. (1994). “Stock market reactions to announced corporate illegalities.” Journal of Business Ethics 13, 583–613.
D’Avolio, Gene M. (2002). The market for borrowing stock, Journal of Financial Economics 66, 271–306.
Dechow, P. M., Ge, W., Larson, C. R., & Sloan, R. G. (2011). Predicting Material Accounting Misstatements. Contemporary accounting research, 28(1), 17-82.
Diamond, D. and R. Verrecchia (1987) “Constraints on Short Selling and Asset Price Adjustment to Private Information.” Journal of Financial Economics 18, 277-312.
Diether, K.B., K. Lee, and I.M. Werner. (2009) Short-Selling Strategies and Return Predictability, Review of Financial Studies 22, 575-607.
Einhorn, D. (2010). Fooling Some of the People All of the Time, A Long Short (and Now Complete) Story, Updated with New Epilogue. John Wiley & Sons.
Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work*. The journal of Finance, 25(2), 383-417.
Ferris, S. P., & Pritchard, A. C. (2001). Share price reactions to securities fraud class actions under the Private Securities Litigation Reform Act. University of Michigan.
Gande, A., & Lewis, C. M. (2007). Shareholder-initiated class action lawsuits: Shareholder wealth effects and industry spillovers. Journal of Financial and Quantitative Analysis, 44(04), 823-850.
Goldstein, I., & Guembel, A. (2008). Manipulation and the allocational role of prices. The Review of Economic Studies, 75(1), 133-164.
Henry, T. R., & Koski, J. L. (2010). Short selling around seasoned equity offerings. Review of Financial Studies, 23(12), 4389-4418.
Johnson, M. F., Nelson, K. K., & Pritchard, A. C. (2007). Do the merits matter more? The impact of the Private Securities Litigation Reform Act. Journal of Law, Economics, and Organization, 23(3), 627-652.
Karpoff, J. M., Koester, A., Lee, D. S., & Martin, G. S. (2013). Database Challenges in Financial Misconduct Research. Georgetown McDonough School of Business Research Paper, (2012-15).
Karpoff, J. M., and X. Lou. (2010) “Short Sellers and Financial Misconduct.” Journal of Finance 65, 1879–913.
Safieddine, A., & Wilhelm, W. J. (1996). An Empirical Investigation of Short‐Selling Activity Prior to Seasoned Equity Offerings. The Journal of Finance, 51(2), 729-749.
Lamar, M., (2013, October 16). Prince Frog Shares Plummet After Allegations By Short-Seller. Wall Street Journal. Retrieved from http://blogs.wsj.com/moneybeat/2013/10/16/prince-frog- shares-plummet-after-short-sellers-allegations/.
25 Woolridge, J.R., and Dickinson, A. (1994). Short selling and common share prices. Financial
