A STUDY ON WITHDRAWN MERGER PROPOSALS INVOLVING PRIVATE TARGET

ANNCAR Announced cumulative abnormal returnBIDDERCASH Bidder's cash level BIDDERDEBT Bidder's cash level FINCRISIS Financial crisis FDI Foreign direct investment GPD Gross domestic produ

FOREIGN TRADE UNIVERSITY FACULTY OF BUSINESS ADMINISTRATION

-*** -GRADUATION THESIS Major: International Business Administration

A STUDY ON WITHDRAWN MERGER PROPOSALS

INVOLVING PRIVATE TARGET

TABLE OF CONTENTS

ANNCAR Announced cumulative abnormal return

BIDDERCASH Bidder's cash level

BIDDERDEBT Bidder's cash level

FINCRISIS Financial crisis

FDI Foreign direct investment

GPD Gross domestic product

PRIV Private target

PRIVSTOCK Private target and the intended method of payment for the

merger is stock

RELATED Bidder and target are in related industries

RESIZE Relative size of bidder over target

ROA Return on assets

VIF Variance inflation factor

WITHCAR Withdrawn cumulative abnormal return

Table 3.1: Statistical descriptions of variables using in models 22

Table 3.2: Correlation matrix for variables with event window (0, +1) 23

Table 3.3: Results of multicollinearity test for sample event window (0,+1) 25

Table 4.1: Mean cumulative abnormal returns of proposal announcements 29

Table 4.2: Mean cumulative abnormal returns of proposal withdrawals 30

Table 4.3: MHW test for withdrawn abnormal return with event window (0,+1) 31

Table 4.4: Non-parametric T-test for withdrawn abnormal return with event window (0,+1) 32

Table 4.5: Bidder’s valuation effects based upon target public status and 33

Table 4.6: Multivariate analysis result of model 1 35

Table 4.7: Multivariate analysis result of model 2 36

Table 4.8: Multivariate analysis result of model 3 37

Table 4.9: Multivariate analysis result of model 4 38

Table 4.10: Multivariate analysis result of model 5 39

Table 4.11: Multivariate analysis result of model 6 40

Table 4.12: Summary of multivariate analysis results of six models for event window (0,+1) 41

Table 4.13: Multivariate analysis result of model 1 46

Table 4.14: Multivariate analysis result of model 2 47

Table 4.15: Multivariate analysis result of model 3 48

Table 4.16: Multivariate analysis result of model 4 49

Table 4.17: Multivariate analysis result of model 5 50

Table 4.18: Multivariate analysis result of model 6 51

Table 4.19: Summary of multivariate analysis results of six models for event window (-1,+1) 52

Table 4.20: Multivariate analysis result of model 1 55

Table 4.21: Multivariate analysis result of model 2 56

Table 4.22: Multivariate analysis result of model 3 57

Table 4.23: Multivariate analysis result of model 4 58

Table 4.24: Multivariate analysis result of model 5 59

Table 4.26: Summary of multivariate analysis results of six models for event

window (-2,+1) 61

I cannot express enough thanks to my thesis supervisor – Associate Prof., Dr.Nguyen Thu Thuy – for her continued support and encouragement A complicatedquantitative study like this thesis is really a huge challenge in which I have to admitthat sometimes I felt lost and demotivated However, with just a few short talks,Mrs Thuy could clearly guide me the strategy to complete the thesis, helped me onhow to tackle problems, and really inspired me to strive for my best for the study Iwas also surprised by her professionalism and enthusiasm while I submitted mythesis draft in the evening and could receive so detailed and essential comments for

my thesis in just the morning next day It is my luck to have an opportunity to workwith her, in which I can not only complete a good thesis but also learn much fromher professionalism and her research skill

I also would like to express my most sincere gratitude to Dr Cao Dinh Kien, for allhis support and advices If it would not be Dr Kien, I could not approach a widerange of highly quality papers and database, and find the idea for this thesis Inaddition, thanks to the great knowledge and skills of Dr Kien about runningqualitative models, I can have lots of advices which are crucial for my thesis

I am also indebted my friends with all their helps in this work For Nguyen VietDuong, I would like to thank him very much for his sharing about the experience indoing thesis and for guiding and reminding me the basics about statistics andeconometrics For Hoang Ngoc Anh and Nguyen Duc Hung, thanks to their help, Ican access to the database of California State University Fullerton and of La TrobeUniversity, which are the critical condition to complete my thesis

And it does not come to this day by chance, but for all past four years, I owe specialthanks to my lecturers and friends at Foreign Trade University for everything theyhave done for me And foremost, to my family, words cannot express mygratefulness For this and more, this is my gift to them

Truong Huy Hoang

CHAPTER 1: INTRODUCTION

1.1 Background

The phenomenon of mergers and acquisitions has developed to become a highlypopular form of corporate development to create growth and diversity (Cartwrightand Schoenberg, 2006) Merger and acquisition are a vital part of both healthy andweak economies and are often the primary way in which companies are able toprovide returns to their investors, stakeholders, and owners (Sherman, 2010)

However, in general, out of ten proposals for a merger in Australian StockExchange, one of them will be withdrawn In the world as a whole, proposals thatare withdrawn constitute a ratio of one over twenty (summarized from ThomsonFinancial SDC Platinum™ database) Because of its large proportion in thepopulation of total merger proposals, the withdrawn merger proposals shouldaccount for an important part of academic research in merger and acquisition fieldand also in real life business practices A withdrawn proposal is intriguing as it canreverse previous effects caused by the results from the announcement of theproposal By how much important the effects of an announcement to firm valuation,

we would expect that much important the effects of a withdrawn proposal forinvestors in the valuation of firm value But the sad fact is that many researchershave been focusing on examining the effects of the announcement of a proposal butnot many of them pay proper attention on effects of withdrawn merger proposal

In consideration of a research in merger and acquisition field, it is widely knownthat effects of an announcement of a proposal from a public bidder can vary bymany characteristics, such as those of bidders, targets, market, and from theproposal itself Therefore, it would be expected that the signal resulting from awithdrawn proposal would also be affected by that many attributes A withdrawnmerger proposal requires more thorough and more attentive dedication in examiningwhat influences its variations

In particular, there is an important research aspect, in which we try to examine thevaluation of a bidder in response to a merger bid may be conditioned on whether itscorresponding target is a privately-held or is a publicly-traded company Effect

causing by whether the target is public or private company, which I call it as targetstatus, in firm valuation is expected to be significant as private and public targetsare inherently different because of information asymmetry, and in general, acquirerswill have different ownership implications for takeover strategy for private targetsversus for public targets In other word, the signal relayed from the withdrawal ofmerger bids for private targets may be unique in comparison with those of theirpublic counterparts Previous studies generally ignored merger proposals involvingprivate targets or did not put proper attention to this unique characteristic Given theimplications might have in firm valuation, this thesis contributes an effort toilluminate whether the characteristic of a firm status affects firm value

Researching on the effect of firm status on merger deal abnormal return is importantfor both academics and business practitioners For academic researchers, thisstudies dig into a new corner of merger and acquisition field, which is withdrawalsinvolving private targets, in which helps to enrich the theoretical framework andmight offer opportunity for further exploration One aspect of informationasymmetry, which represents by whether the firm status is public or private, isfurther examined through the effect of withdrawn merger proposals In addition, theeffects of other characteristics previously pointed out by other researchers thataffect firm valuation in a deal are now more strengthened with evidences from thisstudy In practical business life, the implications from this study can provideinsights and useful knowledge for investors and merger consultants Investors canhave a better approach in understanding how valuation of a public firm is differentfrom a private one in a deal Based on this, they can offer fair price between targetand bidder, which is one of the crucial factors contributing to the success of a deal.Additionally, understanding of impact of other characteristics about the deal, such

as the method of payment and effect of financial crisis, is also useful in the process

of making a successful deal

In addition, an important noticeable point here is that not many researches on thetopic of withdrawals of mergers involving private targets have been done in Asiacountries context but there are only very few of them have been done for the UnitedStates In other words, as there are significant differences among above countries’

business environments, there is a doubt that whether existing theories are valid toexplain the effect of firm status, in which a firm is a public or private company, on awithdrawn merger proposal Realizing the lack of empirical evidences in Asiacontext, the objective of this study is to examine how firm status and other controlvariables impact firm valuation from withdrawn merger proposals for selected listedcompanies in Australian Stock Exchange If the empirical evidence supports thegeneral literature for Australia case, we would further examine this effect for othercountries in Southeast Asia and East Asia, and aim to conclude a universalimplication for Asia countries context Due to the lack of time and data, this thesisfirst aims only Australian context but opens for further researches in differentcontexts in the future

This thesis, which is named “A study on withdrawn merger proposals involving

private target”, contributes to literature by assessing the signal relayed by

withdrawn mergers involving private targets By using event study on estimatingvaluation effects and by using ordinary least square regression models to checkcorrelation between variables, this thesis examines empirical evidence to supporthypotheses, generate implications, and contribute to strengthen the theoreticalframework in merger and acquisition field The findings in this thesis demonstratethat announcement of withdrawals of merger proposals involving private targetsproduce negative and significant valuation effects on the bidders’ stock on average.This result is opposite to what has been found for withdrawals of mergers involvingpublic targets, and verifies that the signal relayed for withdrawals of private targets

is unique Even when controlling for the method of payment, it is proved that thisresult remains constant

1.2 Research questions and research objectives

Under the scope of this study, this thesis aims to answer two research questions.Firstly, “In Australian Stock Exchange, do withdrawals of mergers involving privatetargets have unique response in comparison with those involving public targets?”Secondly, “Is the above effect independent with the method of payment?”

Besides research questions formulation, clearly definition of research objectives isessential before conducting a study, especially quantitative research Formalquantitative research should not begin until the objectives have been clearlydefined Clear objectives aim to explain the purpose of the research in measurableterms and define standards of what the research should accomplish (Zikmund,1997) In order to achieve its purposes, this study has the following objectives:

• Estimate and compare valuation effects of withdrawn mergers which iscontrol for firm status (public versus private);

• Examine whether the above effect is dependent on the method of payment ornot (stock payment only versus cash);

• Analyze the estimation effect by multivariate analysis with two key variableswhich are firm ownership (public or private) and method of payment, andother proper control variables based on literature and previous researches;

• Make recommendations and offer suggestions for further researches

1.3 Scope of the study

The object of this research is studying about withdrawn merger proposals inAustralian Stock Exchange, in which the acquirer is a publically traded company,and the target is either private or public company, not including other types ofcompany such as subsidiaries, joint ventures, and government-owned The scope oftime is for the period of ten years, from 2002 to 2013 Given the data available, 68observations are qualified for the scope of this study

1.4 Research methodology

1.4.1 Method of collecting data

Observations of withdrawn merger proposals are obtained from the ThomsonFinancial SDC Platinum™ database Criteria of qualified observations under thescope of this study are (1) the acquirers are listed in Australian Stock Exchange, (2)the announced merger proposals were withdrawn, and (3) announcement date andwithdrawn date are from 2003 to 2012 The stock prices are obtained fromMorningstar® DatAnalysis Premium Database The market benchmark index is

available in Yahoo Finance Data collection process will be introduced in moredetail in the later section

1.4.2 Method of processing data

There are two key methods of processing data in this thesis The first one is theestimation of valuation effects For this approach, I use event study to calculateabnormal return to estimate valuation effects after the announcement of thewithdrawals Then I compare two sub-samples, one is for private targets, and theother is for public targets, to see if there are any differences in the valuation effects

of the two sub-samples The second approach is using descriptive statistics,correlation, and multiple regression models to run multivariate analysis among keyand control variables Then I will analyze the correlation and coefficient ofvariables to generate proper implications

Chapter 2: Literature review

This chapter reviews relevant literature and researches have been conducted before

to establish theoretical framework for this thesis

Chapter 3: Research design

This chapter presents the research designs with two approaches in more detail,which are the estimation of valuation effects and the multivariate analysis Thissection also introduces data collection method in more detail and provides acomprehensive description of data used in this study

Chapter 4: Empirical results and findings

In this chapter, by using both estimation of valuation effects and multivariateanalysis, I would like to explore the features of explanatory variables and therelationship among variables in six different models Furthermore, this chapter alsointroduces robustness check section to ensure that our findings are consistent overdifferent measures

Chapter 5: Conclusions, recommendations, and limitations

This final chapter presents main conclusion as a summary of previous results andfindings It also introduces some recommendations, limitations of this study, andsuggestions for researches in the future

CHAPTER 2: LITERATURE REVIEW

2.1 Merger and withdrawals of merger

A merger is a transaction in which one corporation (the bidder) secures title to theoutstanding shares or assets of another (the target) (Dodd, 1980) Gaughan (2011)defines mergers as “a combination of two corporations in which only the onecorporation survives and the merged corporation goes out of existence”.Researchers are also familiar with the definition of Sherman (2010), who defines amerger as “a combination of two or more companies in which the assets andliabilities of the selling firm(s) are absorbed by the buying firm Although thebuying firm may be a different organization after the merger, it retains its originalidentity”

Mergers can be classified into three different categories: the vertical integrationmergers; the horizontal mergers; and the diversification merger The first type,vertical mergers are combinations of companies that are symbiotically related Asstated by Scott (2003) and Gaughan (2011), these mergers happen whenorganizations are engaged in related functions but at different stages in theproduction process

Horizontal mergers, the second type, occur when acquirers and targets performingsimilar functions and they want to merge to increase the scale of their operations Itoccurs, for example, when two competitors combine, and economies of scale arerealized (Scott, 2003; Gaughan, 2011)

The third type, diversification, happens when one company acquire one or moreother companies that are neither trading/operating partners nor have the samebusiness lines, but rather operating in different unrelated industries The extremeform of diversification is the conglomerate (Scott, 2003)

The academic study of mergers involves a wide interdisciplinary field of research.Mergers are ever present in the corporate world, and which become an increasinglyimportant part of corporate strategies However, not every attempt to undertake a

merger between two firms is completed; the failure rate in mergers and acquisitionsthat is reported is actually high (Refsnes, 2012).

With regard to withdrawals of merger, there is currently no official definition forthat term till now in the academic research; however, one can understand awithdrawal of merger as the bidders, both voluntarily and involuntarily, terminatetheir offer to purchase the target There are many reasons that lead to thewithdrawals of merger According to Sherman (2010), some of the reasonsattributed for termination of a merger include a lack of adequate planning, an overlyaggressive timetable to closing the deal, management hubris, managementinexperience in merger deal negotiation, or simply too intensive competition fromother bidders A transaction as complex as a merger or an acquisition has manypotential problems and pitfalls; therefore, it is easy to understand that goodplanning, proper time for negotiation, and management experience in a mergingdeal are critical factors to lead a deal to success

2.2 Empirical evidences of related literature about withdrawals of merger

Empirical evidences on topic of withdrawals of mergers are mainly in US context

In his research, Dodd (1980) finds that regardless of whether the proposal issuccessful or cancelled, stockholders of target firms earn positive abnormal returnsfrom the announcement of merger proposals For merger proposals that areeventually cancelled, on average, stockholders of target firms earn significantnegative abnormal returns on the date of the announcement of the termination ofnegotiations As for the side of stockholders of bidder firms, in both successful andwithdrawn merger proposals, there is evidence of negative abnormal returns forbidders over the duration of the proposals

Asquith (1983) and Bradley et al (1983) examine abnormal stock returnsthroughout the entire merger process for both successful and unsuccessful mergerproposals They point out that increases in the probability of a successful merger bidbenefit the stockholders of target firms, and that increases in the probability ofmerger withdrawal negatively affects both target and bidder’s stockholders There isalso evidence that the stock market forecasts probable merger targets in advance of

the merger announcement, therefore, previous studies have underestimated themarket’s reaction to merger bids.

With regard to method of payment, Chang and Suk (1988) find that on average, in

US context, acquirers that offer common stock experience a positive abnormalreturn On the contrary, this observation is not clearly seen when firms offeringcash In other world, the withdrawals of merger transactions that were financed withstock result in positive and significant valuation effects for bidders The results arenot significant when cash or mixed financing was planned

However, there are conflicts in this issue in current literature Sullivan et al (1994)find that the valuation effect of the acquirer is insignificant, regardless of whetherthe intended method of payment was stock or cash Davidson et al (1989) find thatthe valuation effect of the acquirer is negative and significant at the time of thewithdrawal

This thesis focuses on the topic of withdrawn mergers of privately-held targets, and

in Australian context, which are two aspects that were not put much consideration inprevious studies The unique ownership characteristic of the private target allowsfor a special interaction with the bidder, which should relay unique valuation effectsfor the bidders when proposed mergers with these targets are withdrawn

In the next section, I would like to introduce in more detail about the literature andprevious findings with regard to factors that may have influence in the variation ofwithdrawn abnormal returns The following theoretical framework is the foundation

to establish proper models in the multivariate analysis section

2.3 Factors affecting withdrawn merger proposals’ abnormal return

2.3.1 Target firm status

Target firm status decides whether target is a publicly-traded or privately-heldcompany In consideration of a merger, target firm status has significant impact onbidder returns Target firm status has distinctively different implications on bidderreturns because of the following reasons First, as suggested by Fuller et al (2002),private targets are likely to be sold at a discount in comparison with public targets to

compensate for their lack of liquidity Private targets do not enjoy the benefits ofpublicly-trading as public targets; therefore, the ownership of a private target is noteasily transferable as public one is The lack of liquidity helps bidder to purchasethe target firm at a lower price to remove the disadvantage of liquidity deficiencyonce the target is under ownership of the bidder.

Second, private targets are also different from public targets because they are notrequired to disclose public information This makes the targets less attractive astheir financial information and their intention for a merger is not available, hence,they might be ignored by many prospective bidders Even when a bidder makes theeffort to pursue a private target, there is substantial information asymmetry whichwould make the valuation of target firm become harder, leading to the demand of adiscount for bidder price (Officer et al., 2009)

Combining the above two factors, bidders for private target can demand adiscounted price for the deal; therefore, the mergers involving private targets wouldhave positive returns for bidders in comparison with those involving publiccounterparts The withdrawal of the mergers involving private targets may eliminatethe potential benefits might be generated and therefore affects negatively on bidderwithdrawal abnormal return

2.3.2 Method of payment

The interpretation of a withdrawn merger bid is different when involving privatetargets For merger transactions that were financed with stock, Madura and Ngo(2012) state that the use of stock to acquire a private target relays a favorable signal.Consequently the termination of that merger may eliminate that favorable signal andresult in the negative withdrawn abnormal return This contends that the method ofpayment signals the intrinsic value of bidders to the market, because the bidder withthe intrinsic value information may choose the payment method benefiting thebidders This hypothesis was indirectly supported by Jensen and Meckling (1976)and Myer and Majluf (1984)

In summary, the withdrawal of a merger proposal involving a private target that was

to be financed with stock removes the implicit endorsement of the bidder by the

private target It also eliminates the monitoring benefits that would have occurredfrom creating a new block-holder, which is the private target owner (Madura andNgo, 2012).

2.3.3 Bidder cash level

Bidder cash level should have significant impacts on firm merger decision Cashsituation of bidders influences bidders’ decision on which to use cash or stock aspayment or medium, as supported by Martin (1996) and Harford (1999) Therefore,

we might expect bidder cash level also affect market interpretation of withdrawnmerger decision When the cash level is already low, a decision of a withdrawnmerger might be more acceptable and does not impact bidder stock return Incontrast, if the withdrawal of merger involving private target cannot be attributed toaffordability, it may signal a negative response in bidder stock price

2.3.4 Bidder leverage level

A merger can significantly impact the borrowing capacity of the bidder because itdemands the bidder to raise significant funds to purchase the target (Galai andMasulis, 1976; Travlos, 1987) A withdrawn merger of a private target may be moreacceptable to the market if the bidder’s financial leverage is already high.Conversely, if the bidder leverage level is not so high, the market would be unlikely

to attribute the withdrawals of the merger for the reason of financial leverageburden Therefore, the withdrawn merger of a private target may relay a morepronounced negative signal on the bidders’ returns

2.3.5 Abnormal return from merger announcement

The announcement of a merger proposal and the announcement of the withdrawal

of that proposal are two highly correlated opposite events We would expect thewithdrawal of the proposal will reverse any abnormal gain or loss for bidder whichpreviously caused by the announcement of that proposal (Madura and Ngo, 2012)

If the bidder enjoys positive abnormal return when announces the merger proposal,the bidder would suffer negative returns when that proposal is announced to bewithdrawn And vice versa, if the bidder suffers a loss for the announcement of theproposal, it would have a positive return when that proposal is withdrawn

2.3.6 Multiple bidders

When there is strong competition, and in this particular scenario, there are multiplebidders, the withdrawal of a bidder may prevent them from overpaying for thetarget Therefore, the event of a withdrawal is acceptable for the market as thisaction serves shareholder interest by avoiding wealth transferring from bidder totarget As an explanation for this, Walkling and Edminster (1985) argue that bidderstend to suffer hubris and offer a too high premium to pay to the targets to avoidlosing the deals to other bidders The withdrawal by a bidder may be viewedfavorably to the extent of avoiding overpayment, holding other factors constant.However, the impact of multiple bidders is controversy as Schipper andThompson’s results (1983) indicate that it is difficult to identify the market’sperception of an individual acquisition when firms make multiple bids, as part of anannounced acquisition program

2.3.7 Relatedness

According to Morck et al (1990), mergers of unrelated targets tend to be overpaidand do not serve shareholder interests There are three reasons that explain whymanagers might overpay for unrelated targets First, if managers are not properlydiversified themselves, they would diversify their firms to reduce the risk of humancapital even when diversification offers few if any benefits to shareholders (Amihudand Lev, 1981) Second, to assure survival and continuity of the firm whenshareholder wealth maximization dictates shrinkage or liquidation, managers try toenter new line of business (Donaldson and Lorsch, 1983) Third, when poorperformance of the firm threatens a manager’s job, he has an incentive to enter newbusinesses which he is better (Shleifer and Vishny, 1991)

In all these cases, unrelated target mergers destroy shareholder value, and relatedtargets mergers are more likely to serve shareholder interests Therefore, thewithdrawal of a merger with a related target should cause greater disappointment inthe bidder, and a more negative share price response

2.3.8 Relative size of bidder versus target

According to Fuller et al (2002), there are differences in bidder return with regard

to two control factors: one is relative size of bidder versus target and the other istarget firm status (public or private) For public targets, the larger the relative size ofbidder over the target, the more negative impact it has on the acquirer's cumulativeabnormal return In contrast, for private targets, the returns earned by acquirerbecome more positive as relative size of bidder over target increases

2.3.9 Financial crisis

Increasing economic activity has been shown to be positively related to stockfinancing (Marsh, 1982; and Choe, Masulis and Nanda, 1993) Seung and Bang(2011) find that there is a long-run equilibrium relationship between the set ofmacroeconomic variables and merger activity, implying that the macroeconomicfactors play an important role in determining the trend and return of aggregatemerger activity As a result, we would expect that in the time of financial crisis, badfinancial environment would have negative impacts on bidder returns

2.3.10 Bidder’s return on assets

According to Cyree and DeGennaro (2001), they indicate a significantly negativerelation between abnormal returns and acquirer return on assets, which suggests thatbidding firms which have high pre-acquisition ROA experience lower event-periodreturns This negative effect on the announcement abnormal return should bereversed for the withdrawals Therefore, we would expect acquirer’s ROA havepositive impacts on withdrawn merger proposals

CHAPTER 3: RESEARCH DESIGN

In order to answer the research questions, this thesis applies two researchmethodologies: one is using estimation of valuation effects approach, and the other

is using multivariate analysis approach The two approaches both examine theeffects of withdrawals of mergers involving private targets but focus on differentaspects By using two approaches for the same topic, this thesis could gain a morecomprehensive understanding of the issue and strengthen robustness of theconclusions The two approaches are presented in more detail as following

3.1 Estimation of valuation effects

In order to examine if there are any distinctive differences between firm status andits effects on withdrawals of mergers, I calculate and compare cumulative abnormalreturns of two sub-samples: one includes withdrawals involving public companiesonly and the other involves private companies only

I use the market index for all ordinaries shares of Australian Stock Exchange as themarket benchmark for the estimation of valuation effects due to withdrawn mergerproposals The estimation period applied in the calculation is the (-250,-50) daywindow prior to the withdrawal date The valuation effects are estimated for severalevent windows such as (0,+1), (-1,+2), and (-1,+1) days around the withdrawal date

To ensure that the results are peculiar to mergers with private targets, and notattributed to some unknown factors that similarly affect all types of mergers,after having cumulative abnormal returns, I compare the valuation effects fromwithdrawn mergers with private targets to withdrawn mergers with public targetsand assess whether exists statistically significant difference between those two sub-samples

As I am also interested in examining whether method of payment affecting theabove result, I repeat the same above process with the control for method ofpayment for two above sub-samples I aim to check if payment medium changes thevaluation effects resulting from firm status

3.2 Multivariate analysis

3.2.1 Hypotheses

With the implementation of estimation of valuation effects approach, we can justexamine the impact of the method of payment on withdrawn mergers’ cumulativeabnormal returns In consideration that there may be some other factors besidespayment medium that could affect the subsamples of mergers involving private andpublic targets, I also apply a multivariate analysis to a separate sample of mergertargets I would like to propose five hypotheses to test as followings:

Hypothesis 1: Bid involving private targets has negative valuation effects in

response to a withdrawn merger

To the extent that the bidder experiences a valuation gain in response to anannounced merger bid, the gain will be reversed if the bid is withdrawn Thewithdrawal of the merger eliminates the possible benefits of the bidder frompurchasing a private target at a discounted price which is lower than its actual value.Thus, we expect negative valuation effects in response to withdrawn merger bidsinvolving private targets

Hypothesis 2: If stock is the intended method of payment, it will have negative

correlation with firm valuation on effects of withdrawn mergers of private targets

As the use of stock in a deal of a private target experiences positive returns, thewithdrawals of those deals will reverse that favorable signal We would anticipatethat for proposed mergers that are supported with stock, the valuation effects areworse for private targets than public targets

Hypothesis 3: Bidder’s cash level has negative valuation effects on bidder’s

withdrawals of mergers of private targets

For bidder who already has low cash level, we might expect a decision forwithdrawn merger is due to cash unavailability The withdrawal decision will serveshareholder interest by avoiding pushing bidder cash capacity; therefore, thewithdrawn abnormal return will not be negatively impacted Conversely, if the cashlevel is already high and the withdrawal decision cannot be explained by bidder’s

affordability, we expect a negative correlation with withdrawn merger cumulativeabnormal return to reverse the positive impact that has been caused by theannouncement of the proposal

Hypothesis 4: Bidder’s debt level has positive valuation effects on bidder’s

withdrawals of mergers of private targets

High leverage level is a major concern for bidder when choosing whether or not toproceed a deal; therefore, for firms which already have high debts, market is moreacceptable for the withdrawal of the merger Conversely, bidders with low debt level

do not have sympathy from the market by this reason Therefore, we might expectthat high debt leverage would have positive effects on withdrawn abnormal returns

Hypothesis 5: Announcement abnormal return has negative valuation effects on

withdrawn mergers of private targets

Announcement of a merger and the withdrawal of that merger are two oppositeevents, hence a withdrawal of a merger should reverse the benefits or losses whichhave been generated by the announcement of that merger Therefore, the bidder’svaluation effect at the time of the withdrawal announcement should be inverselyrelated to the bidder’s previous bid announcement effect For this reason, we mightexpect a negative correlation between announced merger abnormal return andwithdrawn merger abnormal return

3.2.2 Models

Given the five hypotheses addressed above, I would like to include five respectiveindependent variables in the model to test whether respective null hypotheses can berejected However, as discussed in the literature review section, there are othervariables might affect withdrawn merger abnormal returns Therefore, I also includefive other control variables in the model, which makes the total number ofindependent variables reaching the number of ten

The full variables model used in thesis is as following:

Model 1:

WITHCAR i = β0 + β1PRIVi + β2PRIVSTOCKi + β3BIDDERCASHi

+ β4ANNCARi + β5BIDDERDEBTi + β6MULTIBIDi

+ β7RELATEDi + β8RESIZEi + β9FINCRISISi + β10ROAi + ui

Meaning of each variable in the above model is explained in more detail as below:Dependent variable:

• WITHCAR: is the Cumulative Abnormal Returns (CAR) to the bidders in the(0,+1) days around the announcement of the withdrawal date of the merger Independent variables:

The following independent variables are included in a multivariate model to explainthe variation in the dependent variable and test our hypotheses

• PRIV: is set equal to 1 if the target is private and 0 otherwise A negative andsignificant coefficient of PRIV would support our hypothesis that valuationeffects of withdrawn mergers are worse when they involve private targetsthan public targets

• PRIVSTOCK: is assigned a value of 1 when the proposed merger involves aprivate target and at the same time is to be financed with stock and 0otherwise A negative and significant coefficient of PRIVSTOCK wouldsuggest that for proposed mergers that are supported with stock, the valuationeffects are worse when they involve private targets than public targets

• BIDDERCASH: is measured as the ratio of acquirer’s cash level over totalassets

• BIDDERDEBT: is measured as the ratio of acquirer’s total debt over totalassets

• ANNCAR: is the cumulative abnormal return during the (0,+1) period at thetime of the initial merger bid announcement

The five key variables listed above are included in the model with the main purpose

is to test our hypotheses As discussed in literature review section, we shouldinclude some other characteristics that can impact on withdrawn mergers’

cumulative abnormal returns in the model Based on the literature review section,besides five variables that can affect withdrawn abnormal returns, there are fiveother characteristics we need to consider, and those five characteristics arepresented through five variables as following:

• MULTBID: is assigned a value of 1 if there are multiple bidders and 0otherwise

• RELATED: is a dummy variable set equal to 1 for mergers by parties of thesame two-digit Standard Industrial Classification (SIC) codes and 0otherwise

• RESIZE: is the relative size of total assets of acquirer over the target

• FINCRISIS: is assigned a value of 1 when the time of proposed merger isfrom 2007 to 2010 and 0 otherwise Under the time scope of the study,financial market suffered financial crisis for the above mentioned time

• ROA: is return on assets of the bidder

To the extent that the initial bid effect (ANNCAR) is related to the othercharacteristics that may affect the bidder’s valuation effect at the time ofwithdrawal, such as BIDDERCASH and BIDDERDEBT, I would like to applyalternative reduced-form models that exclude some of the characteristics that mayresult in multicollinearity Five reduced-form models that I use in this study areintroduced below:

Model 2: Reduced-form model

WITHCAR i = β0 + β1PRIVi + β2PRIVSTOCKi + β3MULTIBIDi

+ β4RELATEDi + β5FINCRISISi + ui

Reduced-form model 2 is the model with the least number of variables, in which Iaim to test only two key variables, which are PRIV and PRIVSTOCK, and isolateother side effects as much as possible The impact of abnormal return caused by theannouncement of proposal, which are presented by three variables ANNCAR,BIDDERCASH, and BIDDERDEBT, is not considered as these three variables arehighly correlated and may lead to multicollinearity Two variables, which areRESIZE and ROA, are removed from the model as these two variables have not

been used in previous studies but are just first proposed to test in this thesis.Therefore, I do not include the two newly proposed variables to prevent anyunexpected side effects from them The FINCRISIS variable is also a newlyproposed one; however, I observe that this variable is significant and would be animportant finding in this thesis Therefore, this variable is included in the model sothat when analyzing the results of all six models, I can make conclusion that thisvariable is consistently significant in all models applied in this study.

Model 3: Reduced-form model

WITHCAR i = β0 + β1PRIVi + β2PRIVSTOCKi + β3ANNCARi + β4MULTIBIDi

+ β5RELATEDi + β6RESIZEi + β7FINCRISISi + β8ROAi + ui

In comparison with model 1, reduced-form model 3 does not have two variables,which are BIDDERCASH and BIDDERDEBT As pointed out above, ANNCAR,BIDDERCASH, and BIDDERDEBT are highly correlated; therefore, I isolateBIDDERCASH and BIDDERDEBT to see whether the results are consistent

Model 4: Reduced-form model

WITHCAR i = β

0

+ β1PRIVi + β2PRIVSTOCKi + β3BIDDERCASHi

+ β4BIDDERDEBTi + β5MULTIBIDi + β6RELATEDi

+ β7RESIZEi + β8FINCRISISi + β9ROAi + ui

For reduced-form model 4, in order to examine the possibility of multicollinearity,the variable ANNCAR is removed, and two variables BIDDERCASH andBIDDERDEBT are remained in the models Other variables remain constant

Model 5: Reduced-form model

WITHCAR i = β0 + β1PRIVi + β2PRIVSTOCKi + β3ANNCARi

+ β4BIDDERDEBTi + β5MULTIBIDi + β6RELATEDi

+ β7RESIZEi + β8FINCRISISi + β9ROAi + ui

As discussed in the section above, ANNCAR, BIDDERCASH, and BIDDERDEBTare highly correlated Therefore, in reduced-form model 5, I isolate BIDDERCASH

to see whether the results are consistent

Model 6: Reduced-form model

=

β0 + β1PRIVi + β2PRIVSTOCKi + β3BIDDERCASHi

+ β4ANNCARi + β5MULTIBIDi + β6RELATEDi

+ β7RESIZEi + β8FINCRISISi + β9ROAi + ui

In the reduced-form model 6, BIDDERDEBT is removed from the model to testwhether the results are consistent and free from multicollinearity With six modelsintroduced above, we have enough combination of models to analyze results andmake proper implications

3.3 Data and samples

The withdrawn merger observations are taken from the Thomson Financial SDCPlatinum™ database SDC Platinum™ database is the industry standard forinformation on new issues, M&A, syndicated loans, private equity, project finance,poison pills, and more

The market index benchmark is the market index for all ordinary shares of AustraliaStock Exchange taken from Yahoo Finance This index is available in YahooFinance with the symbol ^AORD and is available for the whole research periodtime, from 2003 to 2012

Historical stock prices of the sample firms are taken from Morningstar®DatAnalysis Premium Database Morningstar® DatAnalysis Premium Database is atrustworthy and reliable database, which delivers a comprehensive current andhistorical picture of Australian Stock Exchange listed and delisted companies Itsextensive corporate data dates back to 1998

The sample selection process is as following First, via the Thomson Financial SDCPlatinum™ database, I identify all mergers that satisfy these criteria: (1) acquirersare listed companies, (2) the proposal announcements made in the 2003 to 2012

period in Australia Stock Exchange; (3) merger status is withdrawn; and (4) targetfirm status is either public or private, not subsidiaries, joint ventures, orgovernment-owned Second, I collect historical stock prices of acquirers in thesamples Only those observations that satisfy the requirement of having enough datapoints to calculate abnormal return for the event window (-250, +3) are retained After following the above process, there are 68 observations satisfy therequirements The description of the sample is introduced in next section.

3.4 Sample description

Table 3.1 shows that over the period from 2003 to 2012, for 68 qualifiedobservations in Australian Stock Exchange, the mean announced abnormal return(ANNCAR) for event window (0,+1), (-1,+1), and (-2,+1) are 3.2%, 3.0%, and5.9%, respectively The mean withdrawn abnormal return (WITHCAR) for eventwindow (0,+1), (-1,+1), and (-2,+1) are -0.9%, -2.2%, and -2.3%, respectively.Given a quick look on this result, it seems that WITHCAR is opposite to ANNCARand this observation is in line with our expectation

Table 3.1: Statistical descriptions of variables using in models

WITH CAR (-2,1)

ANN CAR (-2,1)

WITH CAR (-1,1)

ANN CAR (-1,1)

WITH CAR (0,1)

ANN CAR (0,1)

Mean -0.023 0.059 -0.022 0.030 -0.009 0.032Standard Error 0.021 0.048 0.019 0.021 0.015 0.020Median -0.004 -0.006 -0.001 -0.012 -0.006 -0.010Minimum -0.102 -0.095 -0.112 -0.107 -0.112 -0.097Maximum 0.115 0.121 0.109 0.103 0.105 0.106

No of obs 68 68 68 68 68 68

PRIV PRIVSTOC K MULTIBI D RELATE D FINCRISI S

Mean 0.191 0.118 0.235 0.662 0.603Standard Error 0.048 0.039 0.052 0.058 0.060Median 0.000 0.000 0.000 1.000 1.000Minimum 0.000 0.000 0.000 0.000 0.000Maximum 1.000 1.000 1.000 1.000 1.000

Mean -0.274 5.836 0.182 0.460Standard Error 0.104 1.228 0.023 0.330Median 0.016 3.003 0.181 0.231Minimum -0.545 0.348 0.003 0.029Maximum 0.363 39.268 0.299 1.642

With regard to the explanatory variables, our sample in Australian context is similar

to the sample of Madura and Ngo (2012) for US context The magnitude ofannounced cumulative abnormal return (ANNCAR (0,1) = 3.2%) over the period

2003 to 2012 is quite comparable to that of Madura and Ngo which covers theperiod 1980 to 2006 (ANNCAR (0,+1) = 2.58%) Please find in Table 5 in appendix

for more information regarding other characteristics of our sample

The following table presents the correlation between variables in the six models for event window (0, +1).

Table 3.2: Correlation matrix for variables with event window (0, +1)

WITH CAR

ANN

PRIV STOCK

MULTI

FIN CRISIS ROA RESIZE

BIDDER CASH

BIDDER DEBT

and its independent variables Among the explanatory variables, we observe thatANNCAR has negative correlation (-0.059) with WITHCAR, which meansannounced cumulative abnormal return and withdrawn abnormal return run inopposite direction PRIV has a coefficient with WITHCAR of -0.372, which can beinterpreted as withdrawals of mergers involving private targets have negativeimpact on bidders’ returns

We find that the correlation coefficients are not sufficiently large to causemulticollinearity problems among explanatory variables However, to ensureconfidence in our conclusion, I would like to run multicollinearity test for oursample by calculating variance inflation factor

In multiple regression, the variance inflation factor (VIF) is used as an indicator ofmulticollinearity Computationally, it is defined as the reciprocal of tolerance Thetolerance for each of the models is used in several methods (linear regression,logistic regression, discriminant factorial analysis) as a criterion for filteringvariables, and is computed as (1-R²) The R² of each of the models represents alinear relationship between the dependent variable of the model (the Y) and theexplanatory variables (the Xs) All other things equal, we desire lower levels of VIF,

as higher levels of VIF are known to affect adversely the results associated with amultiple regression analysis Various recommendations for acceptable levels of VIFhave been published in the literature Perhaps most commonly, a value of ten hasbeen recommended as the maximum level of VIF (Hair et al., 1995; Kennedy, 1992;Marquardt, 1970; Neter et al., 1989) Therefore, I would like to use a value of ten as

a benchmark for our multicollinearity test If VIF is less than ten, we can concludethat multicollinearity problems are absent in our models The test result isintroduced in Table 3.3 below:

Variable Toleranc e R 2 VIF Benchmar k

Presence of Multicollinearit

4.1 Overview of Australian economy

As reported by Credit Suisse Global Wealth Report, the economy of Australia is one

of the largest mixed market economies in the world, with a GDP of US$1.525trillion as of 2014 In 2012, Australia was the 12th largest national economy bynominal GDP and the 17th-largest measured by PPP-adjusted GDP, about 1.7% ofthe world economy Australia is the 19th-largest importer and 19th-largest exporter

in the world

Figure 4.1: Gross domestic product of Australia (in US Dollar)

Source: World Bank

According to the World Fact Book, the Australian economy has experiencedcontinuous growth and features low unemployment, contained inflation, very lowpublic debt, and a strong and stable financial system By 2014, Australia hadexperienced more than 20 years of continued economic growth, averaging morethan 3% a year

Source: World Bank

Australia's per-capita GDP is higher than that of the UK, Germany, and France interms of purchasing power parity Per Capita GDP (PPP) Australia is ranked fifth inthe world, according to IMF (2011) Australia's sovereign credit rating is "AAA",higher than the United States of America

Figure 4.3: GDP per capita of Australia (in US Dollar)

Source: World Bank

Inflation has typically been 2 to 3% and the base interest rate 5 to 6% In general,the inflation rate in Australian is lower in compare with the world average, asreported in Figure 4.4 Even in the period 2007 to 2008, when the world had aninflation rate of as high as 9%, inflation rate of Australia still remained at a high of

investors and for developing economics

Figure 4.4: World inflation rate versus Australia inflation rate (in percentage)

Source: World Bank

Australia is one of the world's leading destinations for foreign direct investment(FDI), with total FDI stock growing 6.6 per cent to reach a record AU$507 billion

in 2011, as reported by the Hellenic-Australian Business Council This growthreflects the upturn in global FDI activity since 2010 and Australia's strongcompetitive position in the global economy

Figure 4.5: Key foreign investment in Australia by region/areas of origin

(A$ millions)

Source: Australian Bureau of Statistics

The country's robust economy, strategic location, strong global trade and investmentties, and proven track record of innovation position Australia as an ideal investmentdestination; Australia ranks amongst the top 10 in those projects highlighted by FDIIntelligence and A.T Kearney's 2012 FDI Confidence Index Australia's inward FDIstock has grown by a compound annual rate of 8.5 per cent

4.2.1 Not control for method of payment

The valuation effects of the merger proposal announcement are reported in Table4.1 As can be seen in the table, for announced mergers involving public targets,acquirers experience negative valuation effects, which is in contrast to positivevaluation effects witnessed in announced mergers involving private targets Thisempirical result is in line with previous studies and with literature, which suggestsmergers involving private targets bring higher returns for bidders

Table 4.1: Mean cumulative abnormal returns of proposal announcements

Public targets Private targetsDays N Mean CAR N Mean CAR

abnormal returns experience significant negative returns

Table 4.2: Mean cumulative abnormal returns of proposal withdrawals

Public targets Private targets

In order to test whether the difference between two sub-samples are statisticallysignificant, I apply Mann-Whitney U test The reason of applying Mann–Whitney–Wilcoxon U test in combination with nonparametric t-test is that Mann-Whitney-Wilcoxon U test is more appropriate when dealing with skewed data asnonparametric t-test are sensitive towards outliers or extreme values (Gibbons,1976) Later, as a robustness check, I also apply nonparametric t-test to compare theresults The results of Mann-Whitney-Wilcoxon U test and nonparametric t-test forevent window (0,+1) are as following:

event window (0,+1)

Mann-Whitney-Wilcoxon Test for Two Independent Samples

Withdrawn Abnormal Return Event Window (0,+1)

Significant yes yes

According to the result table, the two sub-samples are statistically different Forboth one tale and two tale testing, the significance is both above 0.1% level Thetable result of nonparametric t-test is presented as following

t-Difference

value

p-critical

t-Significant

One Tail 0.036 3.26 66 0.0009 1.67 yes

Two Tail 0.036 3.26 66 0.0018 2.00 yes

T-TEST: Unequal Variances Alpha 0.05

Standard Error t-statistics Difference p-value t-critical SignificantOne Tail 0.046 2.54 14.60 0.0115 1.75 yes

Two Tail 0.046 2.54 14.60 0.0229 2.13 yes

According to the result table, the two sub-samples are statistically different Forboth one tale and two tale testing, the significance is both above 5% level

The Mann-Whitney-Wilcoxon U test and the nonparametric t-test confirm thestatistically significant difference in abnormal returns of bidders involving publictargets versus private targets These results support the Hypothesis 1 that the marketresponse to the withdrawal announcement is conditioned on whether the target isprivate or public

4.2.2 Control for method of payment

Following the same process above, I repeat the comparison of bidder abnormalreturns when withdrawal announcements involve public targets versus privatetargets while controlling for the method of payment The results are presented inTable 4.5 Panel A of Table 4.5 presents results based on transactions in which cash