Sellerss reactions to takeover rumours and announcements before ma the event study on hose

While many previous studies have found evidence on stock price run-ups and high trading volume in pre-announcement, the controversy on pre-announcement trading in target stocks highly in

UNIVERSITY OF ECONOMICS ERASMUS UNIVERSITY ROTTERDAM

HO CHI MINH CITY INSTITUTE OF SOCIAL STUDIES

VIETNAM THE NETHERLANDS

VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

SELLERS'S REACTIONS TO TAKEOVER RUMOURS AND ANNOUNCEMENTS BEFORE

M&A: THE EVENT STUDY ON HOSE

BY

NGUYEN THI HONG NGOC

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, DECEMBER 2017

UNIVERSITY OF ECONOMICS ERASMUS UNIVERSITY ROTTERDAM

HO CHI MINH CITY INSTITUTE OF SOCIAL STUDIES

VIETNAM THE NETHERLANDS

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

SELLERS'S REACTIONS TO TAKEOVER RUMOURS AND ANNOUNCEMENTS BEFORE

M&A: THE EVENT STUDY ON HOSE

A thesis submitted in partial fulfilment of the requirements for the degree of

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

DISSERTATION DECLARATION

I declares that I am the sole author of "Seller's Reaction to Takeover Rumours and Announcements before M&A: The event study on HOSE", and to the best of my knowledge and belief that all of outcomes included data, ideas and analyzed is written or published another by neither person

Besides, I also declare that my thesis has not been summited or being simultaneous submitted for any Degree, Diploma or any other University or any educational institution

Date: December 01, 2017 Signature:

Nguyen Thi Hong Ngoc

Moreover, I also would like to send my truly sincere my thanks to all of my team "Cao Hoc

Ha Lan" in VNP, who has always supported me, read a lot of my revisions, and given me professional feedback and advice Especially, Mr Khang and Ms Ha provided helpful comments, insights, and also a different perspective to the discussion along the way

Furthermore, I would like to thank the Vietnam - The Netherlands Program for M.A

in Development Economics for imparting me invaluable knowledge and creating a great space for me during my study period I also wish to thank the Doctors who taught me in the M.A environment for giving me an opportunity to study and providing knowledge in the most beautiful university Sincere thanks are also given to the staff of University of Economics Ho Chi Minh City for their help and support to me

Finally, I would like to thank my family, my lover, and close friends for their love and motivating me to higher achievements

Again, many thanks and my honest gratitude to all of you

TABLE OF CONTENTS:

CHAPTER 1: INTRODUCTION 1

1 Problem Statement and Significance Research 1

2 The Research Objectives and Research Questions 3

3 Thesis Structure 3

CHAPTER 2: LITERATURE REVIEW 5

1 Prior Research on Pre-announcement Trading Activity 5

1.1 Developed Market 5

1.2 Emerging Market 6

2 Prior Research on Takeover Rumours 7

CHAPTER 3: HYPOTHESIS DEVELOPMENT 9

1 Abnormal Returns and Abnormal Trading Volume in The Pre-announcement 9

2 Sell-Side Trading Before Announcement Day 9

2.1 Liquidity Trading 9

2.2 Noise Trading 10

2.3 Rational Trading 10

2.4 Testing 11

3 Rumours are reasons why increasing trading volume lead to price movement 11

CHAPTER 4: DATA AND METHODOLOGY 12

1 Sample Data 12

2 Matching Methodology 13

2.1 Selection of Target Firms and Matching Firms 14

3 Differences-in-Differences Estimation 15

3.1 Measuring Abnormal Returns 16

3.2 Measuring Abnormal Trading Activity 16

3.3 Measuring Abnormal Relative Spread 17

3.4 Measuring Abnormal Quoted Depth 18

3.5 Shorting Takeover Rumours - Measuring Abnormal Returns 19

3.6 Robustness Test – Multiple Rumours 20

CHAPTER 5: FINDINGS AND DISCUSSIONS 22

1 Daily Returns and Trading Activity 22

2 Trade Imbalances in Buy-side and Sell-side Transaction 27

3 Change in Ask-Bid Spread and Quoted Depth 30

4 Takeover Rumours: Noise and Rational Traders 33

5 Robustness test – Multiple Rumours 34

6 Relation between Takeover Rumours and Highly Trading Volume-Leading-Returns 35

CHAPTER 6: CONCLUSION 40

REFERENCES 42

APPENDIX 48

CHAPTER 1: INTRODUCTION

1 Problem Statement and Significance Research

During the post-WTO globalization process, Mergers and Acquisitions (M&A) in Vietnam became serious, which might be a complex signal of the economy’s next period of transition (Vuong et al, 2014) In 1990–2010, M&A data in Vietnam showed that foreign firms attempt to acquire domestic ones by M&A about 79.4% Vietnam’s M&A wave happens with anomalies and transactional characteristics, especially it could help explain economic occurrence consisting of security, banking, non-bank financial, portfolio investment, and real estate market (Vuong, 2010) According to Bloomberg (2016), M&A deal activities continue to expand in Vietnam and are looking ahead to hit a fresh record, indicating that M&A models as investment trends are booming strongly in the market Moreover, they also offered a new opportunity for investors or country’s deeper and wider development

The booming of M&A in Asia has caught the attention of both investors and academia (Wright et al, 2005) When corporate acquisition is announced, it is considered as a major new event for trading volume of the target firm While many previous studies have found evidence on stock price run-ups and high trading volume in pre-announcement, the controversy on pre-announcement trading in target stocks highly indicates the non-criminal market anticipation or insider trading

From various kinds of previous literature reviews, most studies on the effects of M&A announcements focus on developed countries There were few studies in emerging countries, especially in Vietnam The investigation of abnormal returns around the day of M&A announcement is analyzed for ten emerging countries in Asia by J Ma, Pagan, & Chu (2009) The findings showed that it has a positive reaction to the announcement of M&A deals Almost all studies aim to investigate abnormal returns in post-announcement that are advantages of active arbitrages These active arbitrages might be against the efficient market and have a certain impact on abnormal returns However, Vietnam’s stock markets have a primary market, different from the previous studies Hence, the investigation of an anomaly, which an abnormally high volume precedes the price movement in Vietnam, is a motivation Moreover, the results from previous researches also showed that increased in both price and trading activity before the announcement Then, there are a lot of debates of whether the pre-announcement trade reflects insider trading or non-criminal market anticipation

Notwithstanding, this study will focus on the abnormally high volume preceding the price movement Based on these findings, it is the motivation to examine the phenomenon of the volume-leading-return with the Vietnam’s dataset and then by explaining its actuality Furthermore, the trading activity translates into the price discovery before the announcement day, or an extreme informational event will be shed light

Compared with corporate activities, the predications of acquisition announcements are so rare However, if investors who can anticipate these acquisition, their profits can be high Thereforee, takeover rumours, often preceded the announcement, may reflect these substantial incentives for investors It is an opportunity to study how the market operates with the highly uncertain information impact on stock prices

Most previous studies examine trading activity on target stock at the time surrounding the announcement period focus on buyer-initiated trades (Gao & Oler, 2012) The price run-

up is due to insider trading assumed by Keown & Pinkerton (1981) Thus, this study will focus on share transactions both buying-active and selling-active which are contradictory

In term of every transaction, it is obvious that it has a buy side and a sell side, but the

“active” for buying or selling is determined by the initiating party The transactions from active sellers or buyers usually are started by posting a market sell or buy order for a transaction immediately Besides, the transaction is facilitating for passive buyers and sellers While most previous research studies used buy-side transactions since they are able to be initiated by potential acquires constructing a toe hold in targets, the seller-side is less clear (Gao & Oler, 2012)

(Vuong, Tran, & Nguyen (2009) showed that using a particular case study, under pressure of unspecified source M&A rumours, Vietnamese investors often quickly spent money on any stock Due to lack of a system for tracking M&A transactions, they just collected a part of the attributes – buy active for their data Thereforee, this study will be motivated to make completely of the lack of data’s Vuong, the sell-side will be collected and examined

Thereforee, using intraday data, this study will show a phenomenon which increased transactions on both active buy-side and active sell-side However, a question asks why sellers sell their target stocks before acquisition announcements Meanwhile, sellers could wait a bit longer and get profits To investigate this problem more deeply, the study follows the previous way, used by Gao in 2008 methodology that evaluated the active “money-losing” selling in target stocks before takeover announcements with three possible explanations

Following the previous studies, this study focuses on the trading activity of target firms in the days before acquisition announcement This study will expand Vuong’ results with larger data for all cases of M&A from 2012 to 2016 with 542 deals at Ho Chi Minh Stock Exchange (HOSE) The abnormal returns and abnormal trading volume are examined

by using M&A announcement data Then, by using Gao & Oler (2012) methodology, the takeover rumour data are collected with 419 rumours between 2012 and 2016 to investigate

“money-losing” sellers

2 The Research Objectives and Research Questions

In particular, the study attempts to answer why sellers sell their target stocks before acquisition announcement There are three main objectives This study uses Zephyr’s data and data stream from 2012 to 2016

The first objective is to find the evidence that abnormally high volume precedes the price movement before announcement date by using intraday data and comparison of trading volume and returns in target stocks Then, the phenomenon of the increase in transactions on both buy-active and sell-active is showed

The second objective is to evaluate possible explanations for the sellers sell their target stocks by divide sellers into three groups: liquidity traders, noise traders and rational traders While liquidity trader is evaluated by expecting increased liquidity follow Lee-Ready algorithm approach, noise traders and rational traders are classified into two kind of investors who get profits or lose money by calculating buy-and-hold abnormal returns around 70 trading days

The third objective is to examine whether rumours are the reasons why high volume precedes the price movement The final objective investigates whether a company has more than one deals, and whether different sellers react in the same ways

The empirical contribution of this study is to utilize the latest data about M&A in Vietnam in 2012 to 2016 to study the trading activity of target firms in the days before acquisition announcement These results provide insight into the M&A environment in Vietnam through stock activations where the market reacts to both certain information (M&A announcements and uncertain information (takeover rumours) With more doubt on the future, the findings also contribute to the kind of investors in Vietnam and policies relevant to Vietnam in term of protecting investors against insider trading and non-criminal market

3 Thesis Structure

The remainder of the paper is organized as follows:

 Chapter 2 presents the literature review

 Chapter 3 develops hypothesis development

 Chapter 4 discusses data, methodology and empirical models

 Chapter 5 illustrates the estimation results, findings and discussion

 Chapter 6 is the conclusion and policy implications

CHAPTER 2: LITERATURE REVIEW

There is a larger number of studies which have examined the influence of corporate restructuring events on stock returns such as returns in the long run and short run, pre-announcements or takeover rumours relating to M&A deals Almost is investigated in US and

UK market and developed countries,

1 Prior Research on Pre-announcement Trading Activity

There are several papers studied about returns and/or volume for target stocks in the period of pre-announcement In the previous studies, most of them showed that it had an abnormal sign about returns and volume for target stocks relative to the pre-announcement day

From subsequent studies, the phenomenon of stock price run-up and abnormally high volume in the prior to the announcement date was found by Jarrell & Poulsen (1989) and Draper & Paudyal (1999) Gao & Oler (2012) also found that abnormally high trading volume precedes significant price movement

With 136 domestic mergers from companies in Japan from 1993 to 2003, the findings around the announcement day show a positive abnormal returns contrasted to US market (Schaik et al, 2004) In the two days pre-announcement, the returns are largest, but lost quickly thereafter It is consistent with other findings that increasing trading volume appears after the announcement

Jensen & Ruback (1983) reviewed 13 the scientific literature on the market for corporate control The evidences are found that the average excess returns are of 30% and 20% for the successful tender offers from target firms’ stockholders, but no abnormal return around the merger

Chronopoulos, Girardone, & Nankervis (2013) examined whether stock markets in the US and Europe Price Efficiency Gains by bank M&A deals Between the stock market’s

reaction and the difference in a bank’s performance before and after a merger find evidence that markets are likely to react differently upon announcement of a bank merger across countries Stocks from targets appear to earn significant abnormal returns in both the US and Europe Particularly, US target shareholders get earn between 1.79% and 6.85% more than their European for 5 days [-2;2] and 11 days [-5;5]

1.2 Emerging Market

In Asian market, abnormal returns for targets stock in developed countries do not statistically significant at convention level By the contrast, in term of M&A deals in Asian emerging market, daily abnormal returns before the announcement day is significantly (Chu

et al, 2009)

Sehgal, Banerjee, & Deisting (2012) examined M&A announcements deals affect stock returns for BRICKS from the period 2005-2009 They found that the significant pre-event returns for 5 out of 6 sample countries, which may be caused by possible leakages in information At the post-event time, while South Africa exhibit significantly positive returns, the strong negative returns are observed in case of India, South Korea and China

An empirical analysis of illegal insider trading by Meulbroek (1992) showed that almost half of the target stock price in the pre-announcement run-up and the abnormal return

on an insider trading day increased average 3% before takeovers occurs on insider trading days Thus, the illegally informed trading is evidenced by some of the pre-trade volume by using court information Song, Zhang, Chu, & Song (2009) discussed that insiders reap benefits via information leakage was indicated from significant daily abnormal returns before the announcement day

Pandey (2001) examines the issue of takeover announcements in India with an empirical investigation of 14 large takeover related open offers and using event study methodology It also found that target firm's stock price and volume run up to the short in announcement

Also in India market, Sehgal et al (2012) conducted a study to see the stock market’s impact to 25 public announcements between January 1, 2004 and May 31, 2005 Although the abnormal returns do not significantly different from zero on the days surrounding the event, the abnormal returns were sustainably negative in both internal and external strategic decisions companies study consisted of that the market penalized merger announcements Interestingly, the means of the post announcement period group were lower than the group means of the pre-announcement period

While some investors are more willing to hold more target stocks, others sell more stocks since more information about financial information of firms involved in the M&A and the terms and conditions of merger proposal made them change their minds (Wang & Wong, 2009) In case of toeholds, it looks like a pay-off between toeholds bidders and other bidders For example, it showed that toehold bidders are more aggressive than other bidders, and they willing to increase the selling price of the target Burkart (1995) and Singh (1998); Huang, Jiang, Lie, & Yang (2014)

Thus, some investors that researchers, analysts, or investors may be able to anticipate the takeover, and their trades impound this anticipation into prices since they use publicly available information

Hence, the active-selling in the pre-announcement should be investigated more There is only study examined active-selling of United State targets before announcement day by Gao

& Oler (2012) In Vietnam market, there is no study examined about this problem

2 Prior Research on Takeover Rumours

Most empirical research on takeover rumours showed price run-ups related to rumours have been observed in developed countries such as United State and United Kingdom

In the short-time, takeover rumours will be cause prices to move in a single direction which is an abnormal trend (Rose (1951) and Kosfeld (2005)) In French market, the short-term impact of takeover rumours on target stock prices was examined in 2016 by Laouiti, Msolli, & Ajina (2016) The takeover rumour data were collected from three sources: news agencies, newspapers, and Web sites in from 1997 to 2012 They found that the takeover rumours information has a significant impact on the prices of target firms around and after the date of rumour appearance Particularly, the target firm performance with the best way around 50 days after rumour appearances had an average returns of 4%

Zivney, Bertin, & Torabzadeh (1996) divided takeover rumours data in two columns that published in the “Heard on the Street” (HOT) column which is a similar approach as Pound

& Zeckhauser (1990) and the "Abreast of the Market" (AOTM) column Regarding to HOT column, these rumours exhibit rapid price stabilization when they were published However,

in term of AOTM column, these rumours are associated with short-term over-reactions Their findings also showed that price run-up during the 20-day period before publication of the rumour

Investor reactions to merger and acquisition rumours was examined through 6,878 rumours from a public firm listed on U.S by Ma & Zhang (2016) Rumours data included 55

rumours turn out to be true in 519 takeover rumours, about 10.6%, while it is about 20.8% of them become official acquisition announcements within three years The findings showed that average cumulative abnormal returns of rumour target firms are 4.78% over the three days around the rumour, but in three months, abnormal returns are -4.48% Meanwhile, the negative returns implied that investors overreact to acquisition rumours Moreover, evidence against investor rationality also found

Yang & Luo (2014) investigated the influence of rumour clarification on stock returns

in the Chinese stock market between 2007 and 2011 under market conditions: bull market and bear market While the average cumulative abnormal return is significantly positive in a bull market after the clarification event, it is significantly negative cumulative abnormal returns in the bear market period Furthermore, in the neutral market period, investors behave more rationally than in the bear or the bull market periods In bull and bear markets, investor behaviors do not react differently in response to the contents of clarification announcements which irrational reaction to clarification announcements is mainly from individual investors Especially, rumours that can credibly predict impending events from rumours that cannot distinguished by investors in Chinese stock market

A subsequent study which expand on those of Zivney et al (1996) was examined by Gao & Oler (2012) They also take a similar approach with two data columns collected, but refine their shorting strategy by considering market condition and firm characteristics The findings showed the positive abnormal returns over the 70 days following the publication of the rumours Moreover, if a majority of takeover rumours failing to materialize into actual announcements, the profits of shorting rumoured targets can be attributed

Against this background, this paper will be focused on the significant pre-announcement trading volume mainly comes from rumoured target firms by link the literatures on pre-announcement trading activity and takeover rumours by showing Meanwhile, the phenomenon of volume-preceding-return observed in the aggregate sample of takeover targets from this study may be driven by rumours Specially, the sample of takeover rumours will be not collected by hand or variety sources as the previous studies, the takeover rumours data will be used from aggregate sample of takeover targets of Zephyr’s data from 2012 to

2016

CHAPTER 3: HYPOTHESIS DEVELOPMENT

1 Abnormal Returns and Abnormal Trading Volume in The announcement

Pre-As the above discussions, most of the research studies showed that in the announcement, it is a sign about abnormal returns and abnormal trading volume, especially in the emerging market Thus, the abnormal returns and abnormal trading volume are expected that highly trading volume will be appeared and created price movements, which is also the first hypothesis

pre-H1: Abnormally high trading volume precedes price movement significantly

2 Sell-Side Trading Before Announcement Day

By investigating more deeply trading volume, the buying-volume and selling-volume

is expected to be unbalance The motivations of active-buyers before announcement day are various, provided from current literatures such as Kosfeld (2005), Song et al (2009)and Laouiti et al (2016); in contrast, it is unclear about motives of active-sellers However, Gao & Oler (2012) suggested that there are three possible explanations: liquidity trading, noise trading, and rational trading could explain active selling in the pre-announcement period The second hypothesis which is the main hypothesis is developed as follows

2.1 Liquidity Trading

The first examination is whether the seller in the pre-announcement represents the clustering of liquidity traders The uninformed liquidity traders were illustrated as naïve investors because of their hype and dump manipulation an equilibrium outcome Ozsoylev, (2008); Chou, Tian, & Yin (2015) Thereforee, traders can be beneficial in the equilibrium from rumours; but uninformed liquidity traders can get loss because of the rumours (Chou et al., 2015) At a time of higher market liquidity, the uninformed liquidity traders who are discretionary could cluster around higher volume stocks in order to trade their stocks (Admati

& Pfleiderer (1988); Gao & Oler (2012) Consequently, the trading volume observed should

be higher and accompanied with improved liquidity in the target shares On the other hands, the target stock spread should rise, and the quoted depth should decline in the pre-announcement period

In this study, the target stock spread uses the bid-ask spread measurements to test whether an increase in trading volume might be associated with liquidity improvement Then, the study uses the check quoted depth and transaction size, another proxy for liquidity, in order to confirm the results

2.2 Noise Trading

The second possible explanations for active-selling is noise trading, meaning that sellers are noise traders who are willing to sell target stocks in the pre-announcement period since sellers believe that these target stocks are overvalued It is suitable for Boubakri, Dionne, & Triki (2008) to say that some target companies are evaluated to be overvalue to the potentials of the transaction, it leads to high prices, and the disbursing does not always yield the anticipated outcome

In the case of Campbell Taggert over the period leading up to Anheuser-Busch’s 1982 tender offer, a large increase in the target volume in the pre-announcement period was found when the effects of identified insider trades were removed by using court records to identify illegal insider transactions (Cornell & Sirri, 1992) Meanwhile, other investors were entering the market and selling at the same time which off-sets informed traders were buying The results also showed that trading by noise traders is consistent with the large rise volume of target company - Campbell Taggert

Noise traders were also called “false informed traders”, who are defined as agents

“fail to recognize the extent of the inside information reflected in the market price and incorrectly believe they have superior information” (Cornell & Sirri, 1992) It is consistent with Chou et al (2015), discussing that while some private information disseminated honestly, some rumourmongers often deliberately added noise since most rumourmongers’ purposes are to trick and control the market by spreading rumours

Consequently, noise traders often lose money because of receiving incorrect information In this study, the noise traders will be evaluated that a trading strategy of shorting rumoured target stocks is losing money or unprofitable

2.3 Rational Trading

Sellers in target stock on takeover rumours also might be rational traders, who bet these takeover rumours may not materialize in an actual announcement (Gao & Oler, 2012)When investors receive a financial rumour, it is extremely important for investors to determine whether rumours convey an honest piece of truthful information which is able to happen in the stock market, or it is a misleading and manipulation such as a false message

Thus, investors should know and how to what rumours affect the expansion of event and the associated financial assets valuation (Chou et al., 2015)

Since Cornell and Sirri, 1992 investigated trading volume for single event – a takeover rumour’Campbell Taggert stock, out of the set of all rumoured takeovers are not considered possibly On the other hands, it is only a portion are able to be actually materialize into public announcements Thus, Chou et al (2015) said that the investor beliefs will update based on the message received and the confidence level of rumours, and then determine the asset price in the market to confirm that rumours are able to materialize in an actual announcement Although, investors receive common information about events or rumours, each investor has different belief and interprets information by the different ways So, it may lead to heavy trading without price movement

Investors who sell the target stock will loss a large money if the rumours given actually materializes into a public announcement Thereforee, investors decisions to short the rumoured target stock trade-off their risk Otherwise, investors who are rational traders will sell target stock if their risk compensation or other compensation source might from from a substantial price correction of rumoured stocks when price run-ups by rumours

Accordingly, the rational traders will be evaluated that a trading strategy of shorting rumoured target stocks will get profit Meanwhile, these takeover rumours may not materialize in an actual announcement

2.4 Testing

As a consequence, to distinguish between liquidity trading, noise trading and rational trading as three possible explanations for selling active, the first thing is using bid-ask spread measure to test whether increased volume is associated with liquidity improvement Then, checking depth and transaction size are tested to confirm the results The liquidity trading is expected both a decrease in spread and an increase in depth in the pre-announcement period The second thing is evaluating whether a trading strategy of shorting rumoured target stocks

is profitable or not If sellers get loss, it is the noise trading In contrast, if sellers get profit, it

is rational trading

H2: Traders who sold stock are liquidity traders and noise traders

3 Rumours are reasons why increasing trading volume lead to price movement

Finally, since the characters of M&A announcement data include certain information (M&A announcement) and uncertain information (Takeover Rumours), the third hypothesis

is that the increasing trading volume creates price movement explained by rumours are expected

H3: Rumours can explain why highly trading volume lead to price movement

CHAPTER 4: DATA AND METHODOLOGY

Zephyr covers acquisitions data based on a variety of reputable sources includes deals status (Completed and Rumour- Expired) and announced date From collected data in the period of 2012-2016, it recorded 542 acquisitions deals selected from listed firms in Ho Chi Minh Stock Exchange (HOSE) Furthermore, firms whose common stocks trading less than

10000 VND per share excluded in order to avoid firms that are in distress Consequence, there are 134 target firms from 542 deals with 26.87 % firms have one deals, firms have more than one deals is 73.13% It is so interesting that there are 6.64% one deals and 93.36% more than one deals compare to 542 deals

To examine the study’s purposes, the data sample will be divided two samples The first sample data included firms have acquisition announcement from Zephyr There are 542 deals with particular announcement date for trading activities in pre-announcement day Specially, the announcement data includes deals which have rumours and non-rumours

Besides, the second sample data is created with firms have takeover rumours recorded from Zephyr with 419 rumours However, because of lack of data, there are 360 rumours examined Gao & Oler (2012) collected rumours punished through Wall Street Journal and used Factiva database to collected takeover rumours with terms: “takeover,” “acquisition,”

“acquirer,” or “target” by hand

Chou, Tian, & Yin (2010) used Thomson Financial SDC Platinum database source data to collected take rumours, and using the date of takeover rumours by using Wall Street Journal and Zephyr record to insure takeover rumours from SDC

Ma & Zhang (2016) classify their rumour data through three ways The first way which they excluded rumours does not unidentified by GVKEY Then, each rumour M&A was classified into five forms: acquisition of minority interest, acquisition of assets, or

acquisition of subsidiaries, merger, acquisition of majority interest, however, mergers and acquisitions of majority interest was focused With the first record of each unique rumours, they were kept and excluded concerning rumour development Finally, merging rumour targets sample with the CRSP and Compustat databases

In this study, in term of profitable strategy, takeover rumour data is different from previous researches Thus, the rumour data does not classify and merge in the same way Ma

& Zhang (2016) method, or collected from other sources in like manner as Gao & Oler (2012) method It might be a gap about data when the takeover rumours in this study are collected solely from a Zephyr record However, it is obvious that rumours about mergers and acquisitions, including information about listed firms at HOSE market as takeover targets

or acquisitions of other companies

In order to analysis the abnormal returns and the trading volume, the matching firms will be matched to target firms through the same 3-digit Standard Industry Classification (SIC 3-digits) and market capitalization (firm size) Therefore, SIC 3-digits collected by Orbis database Since matching firms is used to compare the different abnormal returns and volume calculated

Obviously, the firms in target firms (control group) and matching firms (un-control group) have available financial statement data at least 30 days in the pre-announcement day The intraday data is obtained, including trading price, volume, market price (VN-index), from Thomson Reuter in 2012-2016 While trading volume and trading price are collected from the Thomson Reuter, transaction volume both buy-side and sell-side are collected from Vietstock database

2 Matching Methodology

In order to examine the changes in price and volume of target stocks around acquisition announcements, matching firm is often used Abnormal returns or abnormal volume of a matching firm is defined as returns or volume of a sample firm minus the returns

of volume of its control firm over the same period The control firms are obtained by a matching methodology which pairs target firms and non-target firms on the basis of some observable factors The confounding market- and industry-wide effect should control since takeover activities performance time-series clustering and industry-level variation Moreover, the firm size factors also are control because it might be impact on both trading volume and returns by using matching firms On the other hand, with the same industry (SIC 3-digits), it

is likely to be subject to the same industry conditions for two firms, matching industry can

isolate any industry-specific factors that affect changes in price or volume of sample firms Thus, these factors which are control firms are selected by matching the industry (SIC) and firms size are used mostly in the financial studies

There are numerous empirical studies that control for industry and size effects to control such as Hasbrouck (1985) and Ritter (1991) and the closest firm size (usually proxy

by the market capitalization) within the same industry for matching firms However, some researches solely use the size to control including Loughran & Ritter (1995) and Brav & Gompers (1997)

While matching based on industry and size tries to control for possible effects specific to certain industries and size is the conventional method, assessing abnormal returns of a stock relative to another also matching based on earnings per share tries to control for the current level of EPS include Basu (1983), Kim (1997), and Chia, Czernkowski, & Loftus (1997) In the recent empirical studies, the book-to-market factor is added as a control variable to matching firm Gao & Oler (2012) Following Fama & French (1993), the market and industry-wide disturbances which are firm size and book-to-market factors can affect returns Since lack of data in Vietnam, after M&A performed, the financial data of firms deleted, database sources just archived price and volume data

The matching methodology is preferable to unsystematic chosen control group because it has less selection bias A non-parametric method such as matching is better than regression-based methods since it avoids specifying the relationship between inputs and outcomes The next advantage of this matching method brings attention to the common support problem, which insinuates that they only compare returns and volume performance between target firms and non-target firms when these groups have otherwise similar characteristics consist of industry (SIC-3 digits), market capitalization

2.1 Selection of Target Firms and Matching Firms

The steps of matching firms are described as the following First, the universe of possible matching firms as all firms in the intersection of HOSE is defined with the financial data have to available in the period of the most recent month-end at least 30 days in the pre-announcement day Then, firms have stock prices under VND 10000 will be excluded, 516 deals are chosen

In term of matching firm (non-target firms), firms which have the same 3-digits SIC code as the target will be chosen, with a market capitalization between 70% and 130% of the

target firm As target firms, matching firms which have stock price above VND 10000 will be chosen

Furthermore, it is obvious that the matching firms which represent in target firm have

to over the same period So, the final numbers of observations are 39 pairs of firms for this matching method

3 Differences-in-Differences Estimation

Difference-in-Differences (DID) is a technique employed in impact evaluation The method requires multiple subpopulations, which are separated into control group and treatment group The outcomes in each group before and after the interference are measured

In order to calculate the time trend change which is unrelated to the interference, the change experienced by treatment group (the group which is subjected to the intervention) is adjusted

by the change experienced the treatment group (the group which is not subjected to the intervention) The underlying assumption is that the time trend in the control group is an adequate proxy for the time trend that would have been occurred in the treatment group in the absence of the policy intervention The different-in-differences matching estimator is the best methods compared with others (Smith & Todd, 2005)

In this study, daily abnormal returns and abnormal trading volume of takeover targets is measured by DID

The common methods for investigating an event study, the matching methodology based on find the first difference, events effects to the target firms over the same period, target firms minus matching firms However, this difference is not enough to evidence, a comparison between normal days and even days is the second difference to satisfied fully

Regarding to analysis an event study, the estimation window could range from 120 days to 210 days follow Campbell, Lo, & MacKinlay (1997) According to Ma et al (2009),

an estimation window was selected 120 trading days [-125; -6] To examined market reactions around M&A pre-announcement deals, (Gao & Oler, 2012) chose 105 trading days [-100; 5] Similarly, this study also applied an estimation window of 105 [−100; 5] trading days to avoid loss of transactions due to the lack of sufficient observations within the estimation window It is similar to calculated daily trading active stock to investigate the buying and selling behaviors in the Korean Stock Exchange by Kim, Kartsaklas, & Karanasos (2005)

3.1 Measuring Abnormal Returns

The daily abnormal returns of takeover target is defined as the difference returns between the target firm and its matching firm (Oler, 2008)

(1) Where:

: The daily returns of the target stocks from HOSE market

: The daily returns of the matching stocks from HOSE market (including all distributions)

t: The day relative to the takeover announcement day

The above abnormal returns measure controls for return effects by using industry, size, and years of the target firm in order to choose matching firm

3.2 Measuring Abnormal Trading Activity

It is different significantly about trading activity between firms and industries, the abnormal trading activity will be determined following method The first thing is a target company's normal trading activity is defined by taking the average daily volume over an estimation period, days [–100; –51] comparative to the acquisition announcement date The abnormal volume at the first step is calculated by taking the daily volume on a given day subtract normal volume, then divide by normal volume level

Where:

: Daily volume of target firms on a given day (t)

: Average daily normal volume of target firms at days [-100; -51]

With this approach, abnormal volume which is called the mean-adjusted abnormal volume considered trading activity of specific firm However, abnormal volume is above calculated not represent for trading volume affected by acquisition announcement event Since, there are numerous contemporaneous events might affect trading activity, but unrelated to the pending acquisition announcement Obviously, those contemporaneous events must be controlled to calculated abnormal volume effected by acquisition announcement exactly Hence, the market- and industry-adjusted abnormal trading volume for target firms is calculated by repeat the first steps for each matching firm, and then less the mean-adjusted abnormal volume of the matching The measure of abnormal trading activity effected by acquisition announcement is as follow:

Where:

: Daily volume of target firms on a given day (t)

: Average daily normal volume of target firms at days [-100; -51]

: Daily volume of matching firms on a given day (t)

: Average daily normal volume of matching firms at days [-100; -51]

3.3 Measuring Abnormal Relative Spread

Relative spread is defined as the difference between the ask and bid prices over the transaction price (Lee, Mucklow, & Ready, (1993); Gao & Oler (2012)) It is calculated by Vietnam Dong spread divided the mid-point of bid-ask price, averaged for stock in day

Where:

: Relative spread at time t

: Buy price (bid price) at time t

: Sell price (ask price) at time t

Abnormal relative spread is measured by daily relative spread calculated above subtracts the mean-adjusted average of daily relative spread of the stock over an estimation period, days [–100; –51] comparative to the acquisition announcement date

Where:

: Relative spread of target firm at time t

: Average relative spread of target firm at days [-100; -51]

with:

: Relative spread of matching firm at time t

: Average relative spread of matching firm at days [-100; -51]

Moreover, the top and bottom of observations will be truncated about 1% because of the influence of outliers

3.4 Measuring Abnormal Quoted Depth

Liquidity is measure by depth which are the number of shares the market maker is willing to buy or sell at the quoted prices (Lee et al (1993);Gao & Oler (2012)) Abnormal quoted depth is defined in the same method as abnormal volume (abnormal buy-side volume

or abnormal sell-side volume) First, each firm 's normal depth is calculated by taking the average daily volume (buy-side or sell-side) over an estimation period, days [–100; –51] comparative to the acquisition announcement date Then, abnormal depth is calculated as the percent deviation of daily quoted depth from its normal depth levels

Where:

: Daily trading volume in buy-side of target firms on a given day (t)

: Average daily normal quoted depth (buy-side) of target firms at days [-100, -51]

Where:

: Daily trading volume in sell-side of target firms on a given day (t)

: Average daily normal quoted depth (sell-side) of target firms at days [-100; -51]

For comparison, each matching firm is calculated similar to each target firm

Where:

: Daily trading volume in buy-side of matching firms on a given day (t)

: Average daily normal quoted depth (buy-side) of matching firms at days [-100, -51]

Where:

: Daily trading volume in sell-side of matching firms on a given day (t)

: Average daily normal quoted depth (sell-side) of matching firms at days [-100; -51]

3.5 Shorting Takeover Rumours - Measuring Abnormal Returns

The takeover rumour is difficult to determine by particular classification, since rumours can begin from various source and take various forms Some rumours can be spread through word of mouth or newsletters by insiders, others can be provided from internet as a new forum for rumours Before the merge and acquisition of firms are announced public, rumours of takeovers often circulate initial Chou et al (2015) provide evidence suggesting that an initial overreaction of the market to the takeover rumours These evidences are consistent with Zivney et al (1996) When a takeover rumour driven price run-up, it is as advantage for some investors who are aware of the overreaction market Therefore, they might be shorting stocks as a profitable strategy However, rumours can be distributed into two types whether they can credibly predict impending events or not, and can be statistically distinguished by returns of rumoured takeover targets before rumours publication (Chou et al, 2010)

Follow published rumours data, the takeover rumours of each firm will be started at the rumour date, and then the short position will be hold for a period of 70 trading days and the long position will be taken at the same time Returns to an implementable hedge portfolio

is denoted on returns in the takeover rumours case The abnormal returns for target rumour is calculated on two ways: cumulative returns (CAR) and buy-hold-returns (BHAR) However, Fama & French (1998) said that in the real life investment experiences, buy-and-hold returns

are a better representation cumulative returns Therefore, holding the short and long position for the full 70 trading days before closing out will be calculated as market adjusted buy-hold-returns (BHAR)

Where:

: Returns of stock i at time t

: Returns of market at time t

Moreover, the takeover rumours from all firms will be divided into two types: all firms and extremely large firms since most extremely larger firms are often more to be acquirers than targets and its takeover rumours is not confident, and less believed to see overreaction clearly The year of takeover sample will be classified on “all-year” and “hot-year” since a real-life investor would have more sensitive measures in the period of “hot” and

“cold” takeover rumours The “hot” takeover years is designated years 2014 to 2016

3.6 Robustness Test – Multiple Rumours

Since M&A market in Vietnam is different from others, there are 96% M&A firms has more one deals in the period of 2012-2016 To investigate the action of traders whether traders will continue or not reaction as at a first rumours Data will be classified gap of time between a first rumours and next rumours

where:

: Dummy variables of firm i at time t = 0;1;2

t: three periods: 0: zero to 6 months; 1: 6 months to 12 months and 2: more than 12 months

Furthermore, the first rumour date begin zero, then the next rumours will be calculated If the firms have three rumours or more, the next rumours as third rumour or fouth rumour will be investigated by compare with the first rumour

Overall, the below frame will show how to investigate the research purposes:

Matching Methodology

Abnormal Trading Volume & Abnormal Returns

Signed Imbalance Trading Volume (Buy- & Sell-side)

Evaluated Seller (Liquidity, Noise or Rational Traders)

Test Rumours explain Trading Volume before Announcement

Use DID Methodology