The effect of short-sale disclosures : empirical study in European markets

(1)

1 / 40

Master Thesis

The Effect of Short-sale Disclosures: Empirical

Study in European Markets

MSc Finance- Assets Management

Faculty of Economics and Business

Amsterdam Business School, University of Amsterdam

Name: Mengyue Li

Student number: 11391200

Supervisor: Dr. L. Zou

Date: 1-7-2017

(2)

2 / 40

Statement of Originality

This document is written by Student Mengyue Li 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)

3 / 40

Abstract

This paper investigates the general effect of short-sale disclosure policy in the European region, the influence of the first disclosure and characteristics of firms on stock returns and follow–on behavior in short selling. The empirical results show that disclosure policy has a significantly negative impact on the stock bid-ask spread and improves the liquidity of equity markets. It also shows that disclosures have a negative effect on stock returns for a longer period after disclosure announcement rather than immediate effect. Moreover, there is evidence supporting the existence of follow-on behavior in short selling activities.

(5)

5 / 40

1. Introduction

Financial market participants are always deeply concerned about short-sale, especially after 2008 financial crisis. Short seller plays acrucial role in the market. Most academic research argues that short selling is necessary for the normal operation of the market and it improves market informational efficiency and reduces market frictions with faster price discovery. On the other hand, the others argue that short-sale will damage the stability of the market, short sellers may manipulate stock price and damage other investors’ benefit, and it may even contribute to the crisis. That is the reason why short-sale is prohibited in many countries now. Regulators mainly use three kinds of policy on short-sale with scrutiny on it: prohibit short-sale, restrict short-sale and require the disclosure of large short positions.

Boehmer, Jones and Zhang (2013) argue that banning short-sale damages the market liquidity; Goetzmann, and Zhu (2007) investigate that short sale promotes market efficiency and to prohibit it will damage price discovery; Saff and Sigurdsson (2010) find stocks with lower short-sale constraints which are measured as high lending supply, have higher price and informativeness.

Judging from the research mentioned above, prohibiting or constraining short-sale both have a negative effect on liquidity and price discovery. Alternatively, regulators began to use another short-sale policy—big short position disclosure. And in recent years, this policy are becoming increasingly popular, many stock exchanges introduced various disclosure requirements for short selling.

According to disclosure policy, regulators aim at increasing the transparency of big short positions held by investors in certain stocks; reducing different kinds of risks with uncovered short-sale; ensuring that the Member States have clear powers to interpose in different situations to reduce financial stability and systemic risks and increasing

(6)

6 / 40

confidence on short-sale; and ensuring co-ordination between Member States and regulators.

With these objectives, European countries have required the immediate public disclosure of the large short positions. For example, the United Kingdom and Spain began to use this policy after the crisis in 2008, and France started using short-sale disclosure policy in 2011. EU Regulation on Short Selling became effective in 2012, more than 20 European countries required disclosures.

It is reasonable to believe that mandatory transparency of net short positions plays an important role in the global financial market, and it has some impacts on certain aspects of the financial market. However, what is the specific effect? What does this policy exactly bring to us?

There are several previous papers that focused on the disclosure policy: Beber and Pagano (2013) study on short selling bans; Duong, Huszar, and Yamada (2015) test the effect of the mandatory short-sale disclosure policy on the Tokyo Stock Exchange; Jones, Reed and Waller (2016) use the disclosure in European countries as the sample to analyze the effect of disclosure policy on the whole market and individual stocks.

In this study, I focus on the short-sales disclosure policy in the European market, and I mainly discuss questions as follows: What happened to stock price and liquidity of stock markets before and after the policy came into effect? Is the stock return responses to disclosures related to the characteristics of firms? Whether there is follow–on behavior in short selling activities among financial market participants?

I use the short-sale disclosure data and stock price data from 11 European countries— the U.K., Spain, France, Austria, Sweden, Belgium, Italy, Finland, Germany, Ireland, and the Netherlands, to investigate these questions.

When examining the general effect of disclosure policy, I use disclosure data from the beginning of policy (different countries have different starting time) to the end of 2013.

(7)

7 / 40

However, distinct from the existing literature, when testing the effect of disclosures and follow-on behavior, the sample extends through the end of 2016.

Firstly, I examine the general effect of disclosure policy using event study and panel data methods and choose three dates of implementation or modification of disclosure requirement as the event dates. I use four independent variables as the measure to test the general effect of disclosure policy: bid-ask spread, Amihud, turnover and volatility. The empirical results show that disclosure policy has a significantly negative impact on the stock bid-ask spread, especially in large size firms. Hence, it is concluded that the disclosure policy will increase the liquidity of stocks.

Secondly, I examine the effect of disclosure on stock returns and use event study to examine. I only use the first disclosure of each stock until the end of 2016. The results show that disclosures have a negative effect on stock returns for a longer period after disclosure announcement rather than immediate effect. And follow-on disclosures announced by market participators are related to the change in returns of stocks.

Then I examine whether the effect of policy is associated with the characteristics of disclosers. Reed and Waller (2016) have analyzed this problem before, and I use the variables constructed by them which measure the characteristics of disclosers. But in this paper, the number of observation is larger and the period is longer. In my sample of short-sale disclosers, the active short sellers are almost all hedge funds and assets management companies. And regression results show that the characteristics of disclosers are not associated with stock returns response to disclosures, which suggests that markets participators should care about the total big short positions rather than an individual short seller.

Lastly, I analyze the follow-on behavior among disclosers in the market. The observations are the full disclosure data until the end of 2016. I use the lagged variables and interaction variables to examine follow-on activities.

(8)

8 / 40

From the regression result, I find it is more likely that a disclosure is followed by another disclosure in one week or one month, and this probability increases if the previous discloser is a large firm or its headquarter is located in financial centers like New York and London.

The rest of this paper is organized as follows: Section 2 provides reviews of relevant research papers and explains the theoretical background. Section 3 explains the main methodologies that are used in this paper. Section 4 provides background of the European disclosure policy and presents the descriptive statistics of short-sale disclosures. Section 5 shows and interprets the empirical regression results. And sections 6 summarizes the conclusion and further discusses this topic.

2. Literature Review

There is much literature that shows the importance of short-sale in the financial market, Boehmer, Jones and Zhang (2013) analyze the effect of short sale ban issued by U.S. Securities and Exchange Commission (SEC) in 2008 after financial crisis. They find that the short selling ban effects large-cap stocks most, whose market quality suffers a severe degradation. And because of lack of competition, the liquidity in the market is worse; Goetzmann, and Zhu (2007) focus on the short-sale policy among 46 countries around the world and collect data both in time-series and cross-sectional. They conclude that short sale promotes market efficiency, and to prohibit it will damage price discovery; Saffi and Sigurdsson (2011) examine the impact of short-sale constraints on distributions of returns and stock price efficiency. They find stocks with lower short-sale constraints which are measured as high lending supply, have higher price and informativeness.

(9)

9 / 40

Though the short-sale is crucial in the market, there is also hidden danger. Most short sellers are informed traders, and they have more information to reach the short-sale decision. By examining the short sales behavior in 913 Nasdaq-listed firms, Christophe, Ferri, and Angel (2004) find that short-sale activities are significantly associated with post-announcement stock returns, and it indicates that short sellers in these firms are well informed. The authors also point out that this finding is a warning signal for financial market regulators. For the stable development of financial market and to promote healthy competition between market participants, the public short-selling disclosures seem necessary; Boehmer, Jones, and Zhang (2008) argue that short sellers are informed traders as well; Engelberg, Reed, and Ringgenberg (2012) try to find short seller’s source of information. They combine the data of short-sale with news release and find that short sellers almost have no chance to take part in news events, but for the whole market, the volume of short selling significantly increases after events, which means the short seller’s information comes from the public, not the private source. Finally, the authors conclude that profits and advantages of short sellers come from their ability to better analyze the market information and their trading skills. Jiang, Peterson and Doran (2014) focus on three events: Lockup expirations of IPO, the 2008 short-sale ban on financial institutions and option introductions, and they want to examine whether the overconfidence of investors make stocks with high idiosyncratic volatility overpriced under short-sale constraints. The result shows that when regulator relaxes short-sale constraints, stock prices experience a greater decrease, and short interest and trading volume of stocks with high idiosyncratic volatility increase more than those with low idiosyncratic volatility. It suggests that short-sale constraints make idiosyncratic volatility overpriced.

Based on the literature above, short sellers are informed and they can fully use the market information. Under this condition, after the disclosure of big short position released, the price of stocks will decline significantly.

Because the number of short sale transactions accounts for a larger proportion of the whole market transactions than before, more and more market regulators realizes the

(10)

10 / 40

importance of short-sale policy. As discussed above, prohibiting short sale will do more harm than good. As for constraints for short-sale, Grullon, Michenaud and Weston(2015) test whether short-selling constraints have an impact on stock prices and change investment and financing decisions by relaxing constraints on the random sample. They find that the stock price decreases when there is an increase of short-selling activity. Furthermore, small size firms in the sample will reduce their investment and equity issues with these low price stocks. Hence, they conclude that short-selling constraints can influence stock price and have a real effect on investment and financing decisions. Feng, Chan and Yang (2017) use the sample of short selling in the Chinese market to test Miller (1977) hypothesis that stock prices can be above fundamental values when short-sale is constrained. They find that during earnings announcement periods, short short-sale constraints make stocks have significantly negative abnormal returns, it is consistent with the theory that short-sale constraints affect pricing efficiency.

These findings above suggest that it is necessary to relax the short sale constraints, and the authors point out that regulator can also reduce the negative consequences by improving information disclosure in emerging markets.

There are some papers that mainly discuss short sale disclosure policies. Duong, Huszar, and Yamada (2015) test the effect of the market-wide mandatory disclosure policy on short selling in the Tokyo market. Their findings are as follows: Firstly, most investors change their strategies in short-sale under the disclosure policy, and the average short selling declines which means institutional investors are likely to reduce their short-sale transactions. Secondly, smaller and riskier stocks which most institutional investors do not prefer are shorted more than before policy issued. Thirdly, investors become more trend-chasing on short-sale, stocks pricing efficiency declines and the volatility increases under the short-sale disclosure policy.

Jones, Reed and Waller (2016) use the disclosure in European countries as the sample to analyze disclosure policy. They find that under the disclosure policy, short-run abnormal returns are insignificantly negative after disclosure, but some evidence shows that this

(11)

11 / 40

policy reduces short interest, and the reduction in short seller’s information improves liquidity in the market. The authors also analyze stocks with rights issues, they find that stock returns of disclosed rights issues are the same as those non-disclosed counterparts, and there is no evidence showing there is manipulative short selling. Furthermore, they find the follow-on behavior in short selling: a disclosure is often followed by one or more other disclosures in a month. The regression is significant when the initial one is a large firm or centrally located. This finding suggests large short sellers are likely to be well informed. But short interest does not significantly increase after the initial disclosure. With these two results, the authors give an explanation that when the first short position is disclosed, follow-on disclosers have an undisclosed short position as well. The findings in this paper contribute a lot to the research field in short-sale disclosure policy, and it also has some enlightenment for equity offerings’ regulatory policy on short selling. Abusive short-sale during secondary equity offerings is still a big problem for regulators.

Like Jones, Reed and Waller (2016), I investigate the follow-on behavior of short sellers as well. There are abundant previous papers focusing on herding behavior. Christie and Huang (1995) first test the herding behavior in the U.S. market and five years later, Chang et al. (2000) analyze herding activities in Asian markets, such as Hong Kong, Taiwan, Japan and South Korea. Simoes and Valente (2015) focus on herding behavior in a small European market, using the data from 2003 to 2011. Moreover, Grinblatt et al. (1995) test whether mutual funds will invest based on their historical returns and whether mutual fund managers have herding behavior. The authors find evidence that funds managers will sell and buy the same stocks at the same time, which means herding activities exist in funds.

Sias (2004) uses a new method to test the institutional herding behavior by examining the cross-sectional temporal dependence in institutional demand. The author finds that the demand for a stock of institutional investors this quarter is correlated to their demand last quarter and their demand is strongly correlated to lag institutional demand rather than lag returns. Moreover, there is evidence showing that herding behavior

(12)

12 / 40

declines over time and different investors have different herding behavior. These results indicate that investors can infer information through others’ trading, and then herding behavior appears.

Zheng, Li and Zhu (2015) examine the impact of institutional herding on the future excess stock returns in Chinese stock market. They find that future excess stock returns are positively related to the herding behavior measure on both long and short term. And the effect of herding behavior on stocks depends on deferent stock characteristics: the effect is stronger and lasts for a shorter time on larger, liquid or value stocks while the effect lasts for a longer time on smaller, illiquid or growth stocks. Moreover, persistent herd behavior is negatively associated with excess stock returns for a long period but are positively associated with excess stock returns for short to medium time periods. Also, I use event study to analyze the effect of short-sale disclosure policy. MacKinlay (1997) explains how to measure the impacts of an economic event on the value of firms. Using market data, the author measures the effects of an event on a firm’s value by one particular type of disclosure—quarterly earnings announcements. He argues that the effects of an event will be reflected immediately in stock prices. Thus to examine an event’s economic impact, we can use stock prices observed over a relatively short period as an indicator.

Although there is plentiful literature trying to identify the real effect of short-sale disclosure policy, no consistent conclusions have been reached, and it is still not clear whether the short-sale disclosure policy is a net benefit for the financial markets or not. My research intends to continue analyzing this effect and the follow-on behavior around disclosures. Based on the previous literature, my hypotheses are as follows:

After issuing the disclosure policy on big short-sale positions, bid-ask spread will decrease, and the liquidity of stocks will increase, which means this policy makes the market more liquid and efficient;

The disclosure policy will have a negative effect on stock return and the effect of policy is associated with the characteristics of firms;

(13)

13 / 40

Follow–on behavior exists in short selling activities among financial market participators.

3. Methodology

The paper aims to investigate the general effect of disclosure policy; the influence of the first disclosure and characteristics of firms on each stock return, and follow–on behavior in short selling activities among financial market participants. Hence the description of the methodology can be organized into three parts as follows.

3.1 General effect of short-sale disclosure policy

To examine the general effect of disclosure policy, I use event study and panel data methods. As discussed above, many European countries introduced various disclosure requirements for big short positions on different dates, such as the U.K., Spain and France.

In Jones, Reed and Waller (2016), the authors use four distinct event dates: September 22, 2008 for Spanish financials; June 10, 2010 for Spanish market as well; February 1, 2011 for France and November 1, 2012 for the other EU countries.

However, because of the financial crisis in 2008, using the first event date may cause the confounding effect. It is hard to explain whether the abnormal return of stocks is due to the short-sale policy or the stock market is still recovering from the crisis. Hence, I exclude the date of September 22, 2008, and choose the other three dates of implementation or modification of disclosure requirement as the event date. Then, to

(14)

14 / 40

investigate the change before and after issuing disclosure requirement, I create the group of control firms, these firms are that are unaffected by the disclosure policy, and I find data of these firms from WRDS Compustat. The treatment firms are those who subject to the disclosure policy change in Europe. Thus, I can obtain the effect of the policy by making the difference between pre- and post-disclosure.

Next, I use the regression model in Jones, Reed and Waller (2016):

Where

is outcome variable for firm i in industry j and country k either before (t =0) or

after (t =1) event s.

is a “disclosure” indicator variable equal to one if and only if a short position

disclosure requirement is in effect for this observation, and is an “after” indicator variable equal to one if and only if t =1 which means the observation is after event s.

and are fixed effects for industry, country, and event respectively.

The interaction coefficient captures the incremental effect of the disclosure policy on the measured outcome variable.

Moreover, for the dependent variable for each stock, I calculate the average number over the 3-month period. Hence, each stock has both pre- and post-disclosure observations.

3.2 Stock return and characteristics of firms

In this part, I use event study method to analyze the impact of disclosure on the stock return. The sample contains the first disclosure of each stock until the end of 2016.

(15)

15 / 40

According to MacKinlay (1997), Jong and Goeij (2011), I choose the market-adjusted return model to calculate abnormal return:

Where is return on security i in period t; is market return in period t.

To calculate cumulative abnormal returns:

To test significance:

Where

The benchmark I used is the stock index for each country, and the data are available on WRDS (World indices by WRDS). According to the formulas above, I calculate the daily abnormal return and CAR in different event windows.

Then, I test whether the stock returns after disclosures are associated with the characteristics of short sellers. Under the inspiration of Reed and Waller (2016), I use the following model. The regression formula is as follows:

Where is the daily abnormal return of stock i in country k in period t;

is the individual variable related to characteristics of disclosures, Reed and Waller

(2016) construct several variables:

AUM: The natural logarithm of the assets under management of disclosers subject to

(16)

16 / 40

PositionSize (subject to 13F filings as well) MoneyCtr: A variable equal to one if the general headquarter of discloser is located in

New York or London and equal to zero otherwise.

Multiple originations: A dummy variable equal to one if the number of disclosures

occurring on the certain time of the event window exceeds one and equal to zero otherwise.

is fixed effect for the country.

3.3 Follow-on behavior

The last part is to test follow-on behavior. I use the full disclosure data until the end of 2016, and I also use the lagged variables and interaction variables to examine follow-on activities. The regression formula is as follows:

Where is the binary variable equal to one if a particular short seller disclosed a big

short position in a certain stock on a given date for the first time(which means the start of a short position) and zero otherwise.

is a binary variable equal to one if a big short position of the stock

was disclosed between day t-i and day t−i−k and equal to zero otherwise.

is the individual variable related to characteristics of disclosures as discussed above.

is an interaction variable.

(17)

17 / 40

4. Data and descriptive statistics

4.1 Disclosure policy in European countries

In Europe, almost all the countries use the same requirement on the big short-sale position disclosure since the Short Selling Regulation (SSR) issued in 2012. Usually, when short position exceeds the threshold (expressed by the fraction of shares issued by the company), or the position changes, the short seller must disclose the full name of short seller holding short positions, name of the issuer of the relevant stocks, how big the position is and disclosure date.

4.1.1 Short Selling Regulation (SSR)

EU Regulation on short selling and some aspects of credit default swaps became effective on 1 November 2012 with the aim of increasing transparency, reducing risks and encouraging the Member States to protect financial stability.

The Regulation stipulates many rules for the short selling of shares and credit default swaps, and there are several especially for short position disclosures.

Under these rules, for all short sellers, the shares they short must have been borrowed or have arranged to borrow, or have an arrangement with a third party; significant net short positions must be reported to the relevant competent authorities (private share notifications) when they are at least 0.2% of share outstanding and every 0.1%

(18)

18 / 40

increment, disclosed to the public (public share notifications) when they at least 0.5% of share outstanding and every 0.1% increment. For sovereign debt, significant net short positions must be reported to the relevant competent authorities when crossing the thresholds that ESMA(European Securities and Markets Authority) published for sovereign issuers.

Before 2012, some EU securities regulators have adopted regulations in their respective markets with the requirement of short-selling disclosures, such as the United Kingdom, France and Spain. They usually use different thresholds and adopted policy on the stocks in financial sectors.

4.1.2 United Kingdom

Before the entire European disclosure regulation being instituted, the U.K. has already used the short position disclosure policy. Because using short selling may increase the possibility of market abuse during rights issues, on 13 Jun 2008, the U.K. Financial Services Authority (FSA) announced that disclosure of significant short positions in stocks which are undergoing rights issues is required from 20 June 2008, and this provisions will be introduced in Code of Market Conduct.1 As the securities regulator in the U.K., from April 2013, The FSA has now become two separate regulatory authorities—— The Financial Conduct Authority (FCA) and The Prudential Regulation Authority (PRA). The FCA regulate the financial industry in the UK. They aim to ensure that the market keeps stable and promotes healthy competition. And the PRA is a part of the Bank of England and responsible for the prudential regulation and supervision of major financial institutions.

Three months later, FSA introduced another new provision to prohibit the active creation or increase of net short positions in financial companies from midnight of 18

1

See FSA press release FSA/PN/057/2008, 13 Jun 2008 “Financial Services Authority introduces disclosure policy for significant short positions in companies undertaking rights issues”

(19)

19 / 40

Sep 2008, and the list of companies whose stocks had been covered was published the next day. The provision would remain in force until 16 January 2009.2

However, because continuing using disclosure policy will reduce the potential for abusive behavior and distortion of the markets, on 14 Jan 2009, FSA announced that its disclosure policy for significant net short positions in the stocks in UK financial sector companies would be extended to 30 June 2009.3

And on 26 Jun 2009, FSA decided to continue extending the disclosure policy for significant net short positions for financial companies without a time limit. Under this policy, significant net short positions must be disclosed when they exceed 0.25% of a company’s shares outstanding, or increases by 0.1% above that.4

Since 1 November 2012, the U.K. Financial Conduct Authority (FCA) has regulated short selling and certain aspects of credit default swaps (CDS) of all the stocks in the UK, under the Short Selling Regulation (SSR).

4.1.3 France

On 19 September 2008, Autorité des marchés financiers (AMF) which is the French securities regulator announced that the net short position that represents 0.25% or more of the capital of one of the companies must be disclosed to the AMF and the market on the following day. And AMF also banned unsecured transactions of short positions in financial stocks.5

On January 31, 2011, AMF introduced a permanent transparency policy of the short sale disclosure for all stocks trading in French regulated market, the policy came into effect on 1 February 2011, and the ban on shorting stocks introduced on 19 September 2008 was allow to lapse. The holder of the significant big short position must report to the

2

See FSA press release FSA/PN/102/2008, 18 Sep 2008 “FSA statement on short positions in financial stocks”

3_{See FSA press release FSA/PN/009/2009, 14 Jan 2009 “FSA confirms extension of short selling disclosure}

policy”

4

See FSA press release FSA/PN/084/2009, 26 Jun 2009 “UK short selling disclosure policy extended”

(20)

20 / 40

AMF before 15:30 on 2 February 2011, and those who have already reported their positions under the policy issued on 19 September 2008 need send a new report as well. Under this policy, if a holder’s net short position equal or greater than 0.5% of the share capital, he must send a report about this position to AMF, then AMF will post it on its website. And the additional thresholds are 0.1%.

Like the U.K., France used Short Selling Regulation (SSR) from November 2012.

4.1.4 Spain

On September 22, 2008, the Spanish regulator Comisión Nacional del Mercado de Valores (CNMV) adopted the short sell disclosure regulation. The Executive Committee of the CNMV stipulated that any short position exceeding 0.25% of the listed stock must be disclosed before 7 pm on that trading day. And increasing or decreasing of positions equal or greater than 0.25% also should be disclosed.

And on 27 may 2010, the Executive Committee of the CNMV changed the former policy. Firstly, the policy was extended to all stocks, not only financial stocks. The thresholds of short sale disclosure were also changed. They are 0.2% and 0.5% for private and public disclosures respectively. Each 0.1% increase or decrease should be reported as well. In November 2012, Spanish short sale disclosure policy was superseded by the EU policy as well.

4.2 Data source

In measuring how the market responds to the disclosure of short positions, I use the disclosure data in the European region. I selected 11 countries to analyze. (The U.K., Spain, France, Austria, Sweden, Belgium, Italy, Finland, Germany, Ireland, and the Netherlands.)

(21)

21 / 40

For different countries, the start time of disclosure data are different because of different disclosure policies before 2012: The United Kingdom began in June 2008, Spain and France started in September 2008 and January 2011 respectively. For the other nine countries, it was on 1 November 2012.

The disclosure data I collected are from those different beginning time to 31 December 2016. During this period, I obtain more than seventy-thousand disclosures of 5902 positions in 1234 disclosed firms. These disclosure data are available on the national competent authorities’ website of each country.

For analyzing the effect of disclosure policy on stocks returns, I use the data of disclosure until the end of 2013, and for descriptive statistics and analyzing follow-on behavior of short sellers, I use the full sample extend through 31 December 2016.

Each short-sell disclosure contains the full name of person or company who are holding short positions, the name of the issuer of the stock, how big the position is and the date of that disclosure.

Except the short-sell disclosure data, the market measures, such as the daily price data of stocks, and some related company fundamental data are retrieved from WRDS. I collect these data from 2008 to 2016.

Since bid-ask spread data is difficult and expensive to find, I adopt an approach put forward by Corwin and Schultz (2012) to estimate the bid-ask spread using daily high and low data. Daily high prices are almost buy orders and daily low price are almost sell orders. So the ratio of high-to-low prices reflects a stock’s bid-ask spread and variance. When this ratio is estimated over two days, the bid-ask spread part does not change, and variance part is twice as big. So I use this method to estimate bid-ask spread.

(22)

22 / 40

4.3.1 Disclosure data example

According to data I collect above, Figure 1 is an example of disclosure data. It plots the stock price of ELECTROCOMPONENTS PLC against short positions held by other firms for the first five months of the U.K. disclosure policy. On November 1, 2012, BlackRock

Investment Management (UK) Limited disclosed a short position in

ELECTROCOMPONENTS PLC of 0.66% of shares outstanding. And 5 days later, S.A.C. Global Investors LLP disclosed a short position of 1.21% of shares outstanding. And Insight Investment Management (Global) Ltd. disclosed a short position of 1.06% on

December 7, 2012. Together, the three short positions account for 2.82% of the total shares outstanding in ELECTROCOMPONENTS PLC in December. Furthermore, the stock price was continuously growing after Insight Investment Management (Global) Ltd. disclosing its short position.

(23)

23 / 40

Figure 1

Cumulative abnormal returns around the short-sale disclosure

The figure plots the cumulative abnormal returns around the short-sale disclosure. The observations are the full sample of short-sale positions for all firms. The event window is in terms of date of the first disclosure for each stock. Abnormal daily return is calculated by the market-adjusted return model, it is equal to the stock’s daily return minus the market stock index return, the benchmark is the stock index for each country, and data are available on WRDS.

4.3.2 Descriptive statistics

Table 1 presents the summary statistics in different countries. In Table 1, there are more than 70 thousand disclosures of 5902 positions on 1234 different securities. It is clear

(24)

24 / 40

that the United Kingdom has the most disclosures and positions. There are more than 30 thousands disclosures, 2497 positions and 508 firms in the U.K. It shows that U.K. has a well-developed and active equity market. Moreover, the total listed firms in the United Kingdom are collected from London Stock Exchange website, and firms in Sweden can be found in the list of Nasdaq Stockholm, I collect the number in the other nine countries from World Bank website. Using the number of total firms, I calculate the ratio of disclosed firms to total firms. For example, the disclosed firms over the sample period account for 25% of the total firms listed in London Stock Exchange.

From the numbers of positions and disclosed firms, I obtain the average number of positions per disclosed firm. For all firms, it is about 4.78, Finland has the largest number of positions per disclosed firm—7.71, while Ireland has the smallest, 1.93.

I also calculate the average disclosures per position. We can see from the table that each position has about ten disclosures. It is 12.34 for the whole sample. Furthermore, I separate disclosures into different types—initiations, decreases in short interest and increases in short interest. It shows that the percentage of disclosures decrease in short interest is quite similar to that of disclosure increase.

As shown in the table, the average length of holding period is 246 days by excluding positions which are still open. Short sellers in the Netherlands hold big short positions for the shortest time, about 168 days while the longest is Germany, about 360 days. And in the United Kingdom, the average length of holding period is 268 days. The table also shows that the average disclosed short position, for all firms the number is 0.95%, and some of the individual short positions are extremely large, the largest disclosure is a short position held by Fresco S.R.L. in the securities issued by HIBU PLC in Austria, it accounts for 18.5% of shares outstanding.

Table 1 also presents some basic statistics about follow-on behavior. It shows that each firm has about 1.72 follow-on disclosures for the full sample. The number of follow-on

(25)

25 / 40

Table 1

Summary Statistics

All AUT BEL DEU ESP FIN FRA UK IRL ITA NLD SWE

Number of Disclosures 72834 553 2608 4084 5058 4169 8595 30875 156 8207 2144 6385

Number of Positions 5902 65 168 299 428 316 688 2497 27 579 202 633

Number of Disclosed firms 1234 21 34 108 63 41 124 508 14 135 40 146

Average number of positions per disclosed firm 4.78 3.10 4.94 2.77 6.79 7.71 5.55 4.92 1.93 4.29 5.05 4.34

Total firms 7894 71 116 531 3480 124 485 2037 40 550 103 357

ratio of disclosed firms to total firms 0.16 0.30 0.29 0.20 0.02 0.33 0.26 0.25 0.35 0.25 0.39 0.41

Average disclosures per positions 12.34 8.51 15.52 13.66 11.82 13.19 12.49 12.36 5.78 14.17 10.61 10.09

percentage initiations 5.71% 11.75% 6.40% 6.78% 8.70% 7.68% 8.34% 8.36% 16.03% 7.75% 9.61% 9.95%

percentage decreases in short interest 30.96% 40.87% 29.52% 41.87% 48.24% 47.09% 45.47% 46.06% 49.36% 44.97% 44.92% 43.66% percentage increases in short interest 31.62% 47.20% 35.01% 50.98% 42.92% 44.85% 46.18% 45.33% 33.97% 47.28% 45.48% 44.71%

Average length of holding period 246.32 241.88 254.55 360.18 267.46 302.63 222.04 267.96 162.88 236.08 167.44 226.45 Average disclosed short position 0.95% 0.96% 1.00% 1.08% 0.95% 1.01% 0.98% 0.97% 0.60% 0.95% 0.99% 0.98% Maximum disclosed short position 18.50% 18.50% 5.13% 4.73% 5.03% 7.90% 8.91% 12.17% 1.23% 6.25% 6.27% 7.16%

# of follow on disclosure 188 19 82 109 107 105 223 781 14 337 83 212

Average number of follow-on disclosures 1.72 0.90 2.41 1.01 1.70 2.56 1.80 1.54 1.00 2.50 2.08 1.45 Average number of trading days to 1st follow-on 6.63 8.45 6.20 6.95 6.22 7.16 5.98 7.15 8.20 4.81 5.92 5.93 Average number of trading days to 2nd follow-on 8.49 11.75 7.91 10.86 9.36 8.55 9.44 9.29 2.00 8.15 7.94 8.16 Percentage of dates with multiple originations 12.41% 8.70% 13.19% 9.50% 12.27% 17.17% 18.13% 13.54% 0.65% 12.77% 18.95% 11.61%

This table shows summary statistics related to the number of disclosed short positions of each country under disclosure policy. Average number of positions per disclosed firm equals to number of positions divided by the number of disclosed firms; Average length of holding period is calculated by excluding the short positions that are still open (excess the regulatory threshold). Number of follow-on disclosures is the number of first short positions announced by other firms in the (0,20)-day window following the first disclosure. Disclosures by industry show proportions of each industry in the full sample of position. Classification criteria comes from SIC Code in WRDS.

(26)

26 / 40

Table 1 (Continued)

Summary Statistics – by Industry

All AUT BEL DEU ESP FIN FRA UK IRL ITA NLD SWE

Disclosures by industry(SIC)

Agriculture, Forestry and Fishing 0.06% - - - 0.21% - - - 0.03%

Construction 3.69% - - 7.03% 14.85% 5.64% 0.55% 4.89% - 8.81% 10.31% 2.27%

Finance, Insurance and Real Estate 7.80% 56.96% 1.19% 9.40% 19.93% 0.55% 1.66% 9.69% 18.59% 24.02% 6.58% 4.42%

Manufacturing 37.33% 38.52% 31.71% 60.80% 18.23% 77.16% 53.00% 22.97% 44.87% 25.66% 19.12% 63.18%

Mining 5.65% 0.90% - - - 1.66% 7.34% 15.41% - 0.06% 13.25% 3.93%

Retail Trade 11.87% - 28.68% 1.57% 3.18% 0.58% 4.08% 17.92% - 7.35% - 3.96%

Services 9.27% - 0.12% 11.63% 34.70% 2.35% 9.05% 18.37% 15.38% 2.16% 16.79% 5.48%

Transportation, Communications, Electric, Gas and

Sanitary service 10.05% 3.62% 7.36% 5.34% 9.11% 2.13% 12.30% 7.14% - 11.22% 11.33% 10.95%

Wholesale Trade 4.46% - 14.30% 1.30% - 9.93% 0.45% 1.84% 1.28% 1.40% - 0.06%

(27)

27 / 40

disclosures in Italy is the biggest among all countries, which is 2.50, and it means that follow-on behavior is more common in Italy. Moreover, the average number of trading days to 1st and 2nd follow-on behavior are both around one week, it suggests that most follow-on activities happened in a short period after a disclosure. Such as the Netherlands, the first follow-on behavior appeared after 5.92 days, and 7.94 days later there was another follow-on disclosure. Furthermore, 12.41% of disclosures occurred` with other disclosures on the same day, which indicates the informed trading. In the Netherlands, the number of multiple short seller is the largest, and the percentage of dates with multiple originations is 18.95%.

The table also shows the disclosures by industry and we can see that the proportion of the financial firms reached 7.8% percent of all the disclosures while this proportion is the highest in Austria. The manufacturing firms account for 37.33% in the full sample, which is the largest part, and as high as 77.16% in Finland.

5. Results

In this section, the regression results are presented, including the general effect of disclosure policy, the influence of the first disclosure and characteristics of firms on each stock return, and follow–on behavior in short selling activities among financial market participants.

(28)

28 / 40

Firstly, to examine the general effect of short-sale disclosure policy, as I mentioned above, I use event study and panel data methods.

I treat three dates of implementation or modification of disclosure requirement as the event date: June 10, 2010—Spain regulators extended disclosure policy to all stocks; February 1, 2011 — France introduced a permanent policy of the short sale disclosure, and November 1, 2012, the date when policy for the other EU countries became effective. The event window is six-month long: three months before the event and three months after.

The result of regression is shown in Table 2. It is organized in groups: Panel A contains the full sample, while Panel B contains firms with small size. There are four measures of the effect of short-sale policy: bid-ask spread of stock price, Amihud, turnover and volatility.

I calculated bid-ask spread using the method proposed by Corwin and Schultz (2012). Columns 1 shows that the coefficient is significant at 1% level in the full sample and not significant in panel B. The results suggest that the disclosure policy has a significantly negative impact on the stock bid-ask spread, especially in large size firms. Hence, disclosure policy is very likely to increase the liquidity of stocks.

Amihud is a stock-level illiquidity measure as the average of the absolute value of daily returns over the daily dollar volume which is constructed by Amihud (2002). Columns 2 shows that the coefficients are significant at 1% level both in panel A and panel B. For the full sample, the disclosure policy reduces liquidity by 27.8% (log). The result indicates that stocks become more illiquid after using disclosure policy.

The findings from columns 1 and 2 are contradictory. However, Amihud (2002) argues that even a small change in either turnover or volatility can influence Amihud greatly. Hence, I add turnover and volatility. The significant negative coefficient of turnover and positive coefficient of volatility suggest that the change of these two variables influences Amihud measure, hence disclosure policy increases the liquidity of stocks.

(29)

29 / 40

Table 2

General effect of disclosure policy

(1) Bid-ask spread (2) Amihud (3) Turnover (4) Volatility

Panel A: Full sample

-0.012*** 0.2781*** -0.0004*** 0.0206***

t-value -2.97 9.84 -2.64 16.45

Adjusted 0.1585 0.2323 0.0432 0.0737

Panel B: Small size firms

-0.0196 0.3543*** -0.0003*** 0.037***

t-value -1.41 7.77 -2.98 12.17

Adjusted 0.0755 0.3853 0.0916 0.1306

This table presents the regression results for the general impact of the disclosure policy on some market measures. The control firms are those who are unaffected by the disclosure policy and can be found in WRDS, the treatment firms are those who are subjected to the disclosure policy in Europe. Observations are the averages of daily observations for three-month before and after the disclosure policy becoming effective. Bid-ask spread is calculated using a method proposed by Corwin and Schultz (2012). Amihud is the natural logarithm of the average daily stock return scaled by the volume. If the stock return is zero, I treat it as null. Turnover is the trading volume of stocks divided by the total shares outstanding. These data are retrieved from WRDS. The firms in small size group are those in the bottom third of size calculated within the country. Standard errors are clustered at the firm level. The regression contains fixed effects for country, industry and event date. *, **, and *** denote significance level at 10%, 5%, and 1% respectively.

5.2 The effect of disclosure on stock returns

I use event study to analyze the effect of disclosures on stocks returns by using the full sample of stocks. The examining results are shown in Table 3.

In Panel A, I use the first disclosure in each stock to calculate cumulative abnormal return and daily abnormal return with different event windows. The market return (benchmark) I choose is the stock index for each country, and the data are available on WRDS.

(30)

30 / 40

Table 3

Daily abnormal returns and CAR around disclosure

This table shows the cumulative and daily abnormal returns of disclosed stocks in the sample for different event windows. Abnormal daily return is calculated by the market-adjusted return model, and it is equal to the stock’s daily return minus the market stock return, and the benchmark is the stock index for each country. The data are available on WRDS. Panel A contains 1226 first disclosures, while panel B shows the abnormal return in the (0, 1)-day event window. Short-sale disclosures are separated into different types: closeouts (adjust the position under threshold), upticks (increase the short position) and downticks (decrease the short position). Then CAR and abnormal return is calculated following these different types. *, **, and *** denote significance level at 10%, 5%, and 1% respectively.

Seen from A in Table 3, it is obvious that daily abnormal returns are negative and flat in the given event window. For the cumulative abnormal return, the results show that the CARs of stocks are negative in all windows, but only significantly different from zero in a

Panel A: the first disclosed position in each stocks n=1226)

Length of

the window CAR

Daily Abnormal Return (-3,-1)event window 3 -0.0029 -0.0018 (0,1)event window 2 -0.0013 -0.0010 (0,2)event window 3 -0.0011 -0.0008 (0,5)event window 6 -0.0027 -0.0007 (0,10)event window 11 -0.0065 -0.0009 (0,20)event window 21 -0.0155** -0.0011** (0,30)event window 31 -0.0172*** -0.0008*** (0,60)event window 61 -0.0209*** -0.0005*** (0,90)event window 91 -0.0252*** -0.0004***

Panel B: (0,1)-day event

window around events # CAR

Daily Abnormal Return

First disclosed position in each stock 1226 -0.0013 -0.0010 below median size 567 -0.0022 0.0203 at or above median size 659 -0.0007 -0.0001

Closeout 6070 0.0150 0.0017 Upticks 33047 -0.0480*** -0.0021*** Downticks 32749 0.0936*** 0.0017***

(31)

31 / 40

longer event window. The cumulative abnormal return is about -0.13% one day after disclosure, but it is not significant. While 20 days later, CAR is -1.55% which is significant at 5% level. After that, the return is always significant until 90 days after disclosure. Hence, short-sale disclosure effects stock returns in a longer period, about three weeks after the announcement, while it has a little impact immediately after the big short position being disclosed. This conclusion is consistent with Jones, Reed and Waller (2016).

As for Panel B, I use the full sample of short-sale disclosures in the (0,1)-day window. First I divide disclosures into groups by size of disclosed position. The result shows that both cumulative abnormal return and daily abnormal return is insignificantly different from zero, which suggests that the size of disclosed position is not associated with negative sock returns.

Then, I separate short-sale disclosures into different types: closeouts (adjust the position to the level under threshold), upticks (increase the short position) and downticks (decrease the short position). Then I analyze CAR and abnormal return following these different types. The aim of separating is to examine whether the effect of disclosure is different across types. It can be seen that in upticks and downticks samples, coefficients both have significant returns. After big short position disclosure, increasing the short position is likely to decrease in stock return by about 4.8%, while decreasing the position will tend to increase the stock return by 9.3%. This result indicates that increasing the short position by market participators is related to negative returns of stocks while decreasing is related to positive returns.

5.3 The effect of characteristics of disclosers on stock

returns

In each disclosure announcement, we can obtain the information about the short seller. Table 4 shows the top 20 firms who are most active in short selling activities in my full

(32)

32 / 40

Table 4

Twenty Most Active Disclosers

The table presents the top 20 firms who are most active in short selling activities. They have the largest number of big short positions. Number of short positions defined as the total number of big short positions disclosed since the disclosure policy issued in each country. Number of disclosed firms is the number of firms whose securities have been shorted by the discloser. Number of countries is the number of countries in the sample in which the discloser has disclosed short positions. Average short position is the average proportion of stocks outstanding which are shorted by the discloser in a particular position.

sample. We can see that more than half of them are hedge funds and assets management companies. As shown in the table, BlackRock Investment Management (UK)

Limited is the most active short seller, BlackRock, Inc. is a global investment

Discloser

# of short positions # of disclosed firms # of countries Average short position

BlackRock Investment Management (UK) Limited 328 322 9 0.89%

Marshall Wace LLP 314 303 9 0.93%

AQR Capital Management, LLC 186 175 8 1.14%

WorldQuant, LLC 165 165 9 0.80%

JPMorgan Asset Management (UK) Ltd 152 146 7 0.80%

Oxford Asset Management 137 130 8 0.79%

GLG Partners LP 126 122 9 0.76%

BlackRock Institutional Trust Company, National

Association 116 116 8 0.74%

Millennium International Management LP 114 109 8 0.82%

TT International 104 96 10 0.77%

Odey Asset Management LLP 92 89 10 1.61%

GSA Capital Partners LLP 91 88 7 0.70%

Citadel Europe LLP 76 76 9 1.00%

BNP Paribas SA 65 63 10 0.91%

BlueCrest Capital Management Limited 60 60 9 0.66% Highbridge Capital Management LLC 59 58 8 0.91% JPMorgan Asset Management (UK) Limited 58 55 4 0.78%

AKO Capital LLP 49 48 9 1.20%

Numeric Investors LLC 48 45 7 0.75%

(33)

33 / 40

management corporation in New York. It is the largest asset manager in the world and it operates globally in 30 countries. Marshall Wace LLP stays very close to BlackRock. It is a hedge fund located in London. These two firms both have more than three hundred short positions and disclosed firms.

The number of short positions ranges from 47 to 328, while the range of the number of disclosed firms is 47 to 322. Also, Odey Asset Management LLP and GSA Capital Partners

LLP short sell stocks in ten countries, even more than BlackRock and Marshall Wace. It

suggests the globalization and diversification awareness. Odey Asset Management also has the largest average short position, which is 1.61% while the smallest is 0.70%.

Then, I analyze the relationship between characteristics of disclosers and stock returns. As discussed above, under the inspiration of Reed and Waller (2016), I focus on the following variables as measures of disclosers characteristics:

AUM (The natural logarithm of the assets under management of disclosers), PositionSize (dollar value of the disclosed short position divided by AUM multiplied by 10), MoneyCtr (a variable equal to one if the general headquarter of discloser is located in New York or London and equal to zero otherwise) and Multiple originations (a dummy variable equal to one if the number of disclosures occurring on the certain time exceeds one and equal to zero otherwise). I regress abnormal stock return on these variables.

The regression results are shown in Table 5. However, only the coefficient of AUM over the (0,30)-day event window is significant. The results do not support the hypotheses that the characteristics of disclosers are related to stock returns movements response to disclosures announcements. This conclusion has enlightenment to fanatical market participants that there is no need to focus on other disclosers’ characteristics, and the short position held by a single discloser does not have very significant impact on the whole market.

(34)

34 / 40

Table 5

Abnormal returns and characteristics of disclosers

This table presents the several separate regressions of daily abnormal returns on some variables related to the characteristics of disclosers. Each row of the table shows the result of a separate regression. The observations are the first disclosure position of each stock. Abnormal daily return is calculated by the market-adjusted return model, it is equal to the stock’s daily return minus the market stock return, and the benchmark is the stock index for each country. AUM is natural logarithm of the assets under management of disclosers subject to the most recently reported; PositionSize is the dollar value of the disclosed short position divided by AUM multiplied by 10; MoneyCtr expresses a variable equal to one if the general headquarter of discloser is located in New York or London and equal to zero otherwise and Multiple originations is also a dummy variable equal to one if the number of disclosures occurring at a certain time exceeds one and equal to zero otherwise. *, **, and *** denote significance level at 10%, 5%, and 1% respectively.

5.4 Follow–on behavior

Because of the publicity of big short positions, it seems much easier to follow others’ trading behavior in short selling. Hence, I suppose that follow-on behavior exists in short-sale activities.

To test this kind of behavior, I use the logit regression formula which I have mentioned in the previous part of the methodology. Because independent variable is a binary

(0,2) event window (0,10) event window (0,30) event window

SE SE SE

AUM 0.0006 0.0005 0.0000 0.0002 0.0002* 0.0000

AUM above median 0.0014 0.0015 -0.0001 0.0071 0.0005 0.0003

PositionSize -0.1065 0.0765 -0.0220 0.0479 -0.0024 0.0230

PositionSize above median 0.0008 0.0013 0.0000 0.0006 -0.0002 0.0003

MoneyCtr 0.0008 0.0020 -0.0013 0.0009 -0.0004 0.0005

(35)

35 / 40

Table 6

Follow-on positions

The table characterizes the first disclosure of a stock by a given discloser (initiation) related to the previous disclosed positions of the same stock. The observations are the full sample of short-sale positions for all firms. The dependent variable is a binary variable equal to one if a particular short seller disclosed a big short position in a certain stock on a given date for the first time and zero otherwise. is a binary variable equal to one if a big short

position of the stock is disclosed between day t-i and day t−i−k and equal to zero otherwise. This variable combines with the individual variable related to characteristics of disclosures as the interaction variable. AUM is natural logarithm of the assets under management of disclosers subject to the most recently reported; PositionSize is the dollar value of the disclosed short position divided by AUM multiplied by 10; MoneyCtr expresses a variable equal to one if the general headquarter of discloser is located in New York or London and equal to zero otherwise and Multiple originations is also a dummy variable equal to one if the number of disclosures occurring at a certain time exceeds one and equal to zero otherwise. The regression contains the fixed effect for the country, and it is fixed. Standard errors are clustered at the firm level. *, **, and *** denote significance level at 10%, 5%, and 1% respectively.

variable which equals one if a particular short seller disclosed a big short position in a certain stock on a given date for the first time and zero otherwise, in other words, I only use the initial disclosure of each stock to analyze the follow-on behavior. For independent variables, I construct interaction variables by using a binary variable and disclosers’ characteristics. The results are shown in Table 6.

(1) (2) (3) (4) 0.0292*** 0.0597 0.0076 0.0332*** 0.0079*** 0.0408* 0.0070 0.0101*** 0.0122* _-0.0059** 0.4859 -1.0087 0.0194 -0.0105***

(36)

36 / 40

Column1 only contains the lagged dummy variable. The coefficients are significant at 1% level both in (-1,-5)-days event window and (-6, -30)-days window, which means the probability of a new position disclosure on the given day will increase by 2.92% if there is another initial disclosure in last week on the same stock. Similarly, the probability will increase by 0.79% if there is a similar disclosure in the previous month.

Columns 2 to 4 use the interaction variable constructed by using lagged dummy variable and characteristics of disclosers. Column 2 shows that the two coefficients of interaction variable are significant at 10% level, which means if the previous discloser is a big firm, follow-on short-sale behavior is more likely to happen. It can be explained by the fact that high AUM is normally an indicator of the informed trader, and they process better analyzing and trading ability, so other firms prefer to follow them.

As for column 3, the insignificant result suggests that there is no evidence showing that the follow-on short selling is associated with position size of prior disclosure.

In the last column, the coefficient of the interaction variable which is constructed by using lagged dummy variable and location of disclosers in (-6, -30)-days event window is significant at 1% level. The result indicates that if the headquarter of the previous discloser is located in New York or London, the probability of follow-on short-sale increases.

Overall, Table 6 provides the evidence for the existence of follow-on short-sale behavior, especially when the previous discloser is a large firm or its headquarter is located in New York or London.

(37)

37 / 40

6. Conclusion

As short selling plays an important role in the daily operation of the equity market, it is necessary to issue a policy to standardize short-selling. There are mainly three kinds of policies on short-sale which are often used by regulators: prohibit short-sale, restrict short-sale and require the disclosure of large short positions. As some previous literature shows, the first two policies have lots of drawbacks and disclosure policy is widely used in many countries. The United Kingdom and Spain began to require immediate public disclosure of the large short positions policy after the crisis in 2008, and France started in 2011. In 2012, EU Regulation on Short Selling became effective, more than 20 European countries required disclosures.

The aim of requiring big short position disclosures is to improve transparency and monitor financial market, but the effect of short-sale disclosure policy is not all clear until now. And there are some problems that concern market participators and regulators. For instance, what is the effect of disclosure policy on stocks and the whole market? Is the effect of disclosure related to the characteristics of disclosers? Does follow–on behavior in short selling activities exist among financial market participators? In this paper, I investigate the impact of short-sale disclosure policy in 11 European counties. When analyzing the general effect of disclosure policy, the regression results indicate that the disclosure policy has a significantly negative impact on the stock bid-ask spread, which means the policy will increase the liquidity of stocks.

I also use the sample of disclosures data to calculate cumulative abnormal return in different event windows, whose result shows that disclosure of big short position has a negative effect on stock returns in a longer period after disclosure announcement rather than immediate effect. And follow-on disclosures by market participators are related to the change in returns of stocks.

I list the top 20 firms who are most active in short selling activities and find that more than half of them are hedge funds and assets management companies. Furthermore, by

(38)

38 / 40

characterize disclosures and disclosers, I find that the characteristics of disclosers are not related to stock returns movements response to disclosures.

I find the evidence supporting the existence of follow-on short-sale behavior, and the probability of a disclosure followed by another disclosure increases if the previous discloser is a large firm or its headquarter is located in financial centers like New York and London.

However, this research does have limitations. When I use event study to analyze the impact of the policy, there may be some other contemporaneous event, which will interfere the regression, and it is impossible to find all of these events.

The ultimate aim of this policy is to improve transparency of the market and standardize short-selling. And we cannot say that the short-sale disclosure policy is totally good or bad, it has weak points and advantages at the same time. There are still many aspects of the short-selling need to be analyzed further, such as whether the thresholds of big short position size using in Europe is reasonable and how should regulators determine the thresholds.

(39)

39 / 40

Reference

Boehmer, E., Jones, C. M., and Zhang, X. (2013). Shackling short sellers: the 2008 shorting ban. Social Science Electronic Publishing, 26, 1363-1400.

BRIS, A., GOETZMANN, W. N. and ZHU, N. (2007), Efficiency and the Bear: Short Sales and Markets Around the World. The Journal of Finance, 62: 1029–1079.

Boehmer, E., Jones, C. M., and Zhang, X. (2008). Which shorts are informed?. Journal of

Finance, 63, 491–527

Christophe, S. E., Ferri, M. G. and Angel, J. J. (2004), Short-Selling Prior to Earnings Announcements. The Journal of Finance, 59: 1845–1876.

Corwin S A, Schultz P. (2012). A Simple Way to Estimate Bid‐Ask Spreads from Daily High and Low Prices[J]. Journal of Finance, 2012, 67, 719–760.

Christie, W. and Huang, R. (1995) Following the Pied Piper: Do individual returns herd around the market? Financial Analysts Journal, 51, 31-37

Duong, T. X., Huszár, Z. R., and Yamada, T. (2015). The costs and benefits of short sale disclosure. Journal of Banking & Finance, 53(April), 124-139.

Engelberg, J. E., Reed, A. V., and Ringgenberg, M. C. (2012). How are shorts informed? short sellers, news, and information processing. Journal of Financial Economics, 105, 260-278

Grullon, G., Michenaud, S., and Weston, J. P. (2011). The real effects of short-selling constraints. Review of Financial Studies, 28, 1737-1767

Jiang, D., Peterson, D. R., & Doran, J. S. (2014). Short-sale constrains and the idiosyncratic volatility puzzle: an event study approach . Journal of Empirical Finance, 28, 36–59.

Jones, C. M., Reed, A. V., and Waller, W. (2016). Revealing shorts: an examination of large short position disclosures. Review of Financial Studies, 29, 3278-3320.

(40)

40 / 40

Mackinlay, A. C. (1997). Event studies in economics and finance. Journal of Economic

Literature, 35(1), 13-39.

Miller, E. M., 1977, Risk, uncertainty, and divergence of opinion, The Journal of Finance 32, 1151–1168.

Saffi, P. A. C.,and Sigurdsson, K. (2011). Price efficiency and short selling. Review of

Financial Studies, 24, 821-852.

Vieira, E. F. S., & Pereira, M. S. V. (2015). Herding behavior and sentiment: evidence in a small European market. Revista De Contabilidad, 18(1), 78-86.

Zheng, D., Li, H., & Zhu, X. (2015). Herding behavior in institutional investors: evidence from china’s stock market. Journal of Multinational Financial Management, s 32–33, 59-76.

The effect of short-sale disclosures : empirical study in European markets

Master Thesis