The long term effect of corruption on shareholders’ wealth and the
moderating effect of bidder corporate governance in case of
cross-border mergers and acquisitions
Master Thesis International Financial Management
University of Groningen
Faculty of Economics and Business
Author: Vladimir Vasile S2862484 Supervisor: dr. Victoria Purice Co-assessor: Adri de Ridder
Abstract: With cross-border M&A representing the largest part of FDI and considering that previous research mostly overlooks cross-border M&As, the purpose of this paper is to fill this gap and analyze the relationship between bidder’s shareholder value and corruption of the target country in case of cross-border M&As. In order to test the hypothesis that corruption has a negative impact on the acquirer returns, a sample of 193 deals which happened between 2007 and 2014 was used. Furthermore, an additional hypothesis assuming that the bidder’s corporate governance system has a positive moderating role was investigated. In accordance with the expectations, results show evidence for the presumed relationships. However, the hypothesis cannot be confirmed due to the reliability of the evidence found. This study improves the understanding of the relationship between shareholder value and corruption and it encourages further research on this topic.
Table of Contents
1. Introduction ... 1
2. Literature review and hypothesis development... 3
2.1. Mergers and acquisitions ... 3
2.2. Foreign Direct Investment (FDI) ... 4
2.3. Cross-border M&As ... 6
2.4. Corruption ... 9
2.5. Corporate governance ... 12
3. Data and Methodology ... 15
3.1. Sample... 15 3.2. Variable description ... 16 3.2.1. Dependent variable ... 16 3.2.2. Independent variable ... 18 3.2.3. Moderator ... 19 3.2.4. Control variables ... 19 3.3. Model ... 22 3.4. Sample distribution ... 22 4. Empirical results ... 25 4.1.Descriptive statistics ... 25 4.2.Multicollinearity ... 26 4.3.Regression results ... 28 4.4.Robustness test ... 37
5. Discussion and limitations ... 37
5.1.Discussion ... 37
5.2.Limitations and future research ... 38
References ... 39
1
1. Introduction
Change in the actual business environment is mostly characterized by increased
globalization (John et al., 2010) and elimination of borders regarding trade, along with increased deregulation (Danbolt and Maciver, 2012). This has led firms to seek to exploit other markets than their home ones and achieve competitive advantage on a worldwide level (Porter, 1998), creating the perfect setting for cross-border mergers and acquisitions. This economic
phenomenon increased considerably over the past years, fact acknowledged by both scholars, John et al. (2010) reported a ten times increase of cross-border M&As over 7 years, and leading multinational companies which are directly involved (JP Morgan, 2017).
Although the number of cross-border M&As is on the rise, there seems to be a paradox because most of the papers are focusing on the domestic M&As, especially the ones conducted within the borders of US (Erel, Liao and Weisbach, 2012), thus omitting an important part of the deals happening. Therefore the subject of this study is represented by cross-border M&As.
On the other hand, corruption is another phenomenon which gained attention over the last few years and, is of great importance, as losses of approximately 1 trillion $ are yearly recorded due to bribery alone (Weitzel and Berns, 2006). The actuality of corruption can be seen in real life, one recent example being the Romanian protests from February 2017 against a law which aimed to decriminalize corruption felonies. Events like this stress out the importance and gravity of corruption even more. Despite the harsher anti-corruption laws this practice is still present in both the public and private sector (Judge et al., 2011).
Many studies focused on corruption and foreign direct investment (henceforth FDI) but just a few consider them together and fewer investigated the relation between corruption and cross-border M&As (Weitzel and Berns, 2006). This is a gap which needs to be addressed considering that M&As represented approximately 72% of the total FDI volume in 2006 (Malhotra et al.,2010).
Based on these arguments and the identified gap, the following research question is developed:
2 Additionally, the growing importance of corporate governance and the importance of analyzing other governance related variables when studying corruption (Weitzel and Berns, 2006) lead to the formulation of a research sub-question:
“How does the corporate governance of the acquiring company affect the relationship between the value of the bidding firm and the corruption from the target country?”
For answering these research questions, this study resorted to econometrical evidence in order to report on the relationships between corruption and the shareholders’ value of the bidding firm as well as the moderating effect of corporate governance, in a sample consisting of a cross section of countries. The sample includes 193 observations from 25 different countries.
Two hypothesis were developed on the basis of previous literature and a negative relation between corruption and shareholder value was anticipated. Moreover, corporate governance was expected to have a positive moderating effect over this relationship, leading to the acquirer obtaining positive abnormal returns.
3
2. Literature review and hypothesis development
This section presents some of the already existent research and theories which are related to the topic of this paper. As academic literature on M&As and FDI is extensive, only the
relevant findings are reported. In terms of structure, the first part focuses on M&As in general, followed by an overview of the FDI literature, cross-border mergers and acquisitions, corruption and corporate governance. Based on the previous literature, hypothesis are being developed.
2.1. Mergers and acquisitions
The main reason for which firms resort to M&As is the prospect of value creation. This can be achieved through synergy effects that may arise from these activities. In order to be profitable, a M&A should lead to a higher stock price than the added prices of the companies before the merger/acquisition happened. To put it differently, shareholder value creation is achieved by the newly formed or combined entity having a greater value than the former value of the bidder and target summed together without considering the premiums paid (Basuil and Datta, 2015).
Apart from this major goal which is underlying any M&A, every firm can have additional objectives. Walker (2000), developed a set of six different strategic objectives of firms turning to M&As. These objectives were based on primary data such as the opinions of financial analysts and executives of firms taking part in M&As:
The first one is the ambition of geographic expansion through which firms seek to achieve economies of scale by moving to other geographical area being it within the borders of the same country or outside. The latter is known as cross-border M&A and will be discussed in more depth in a following part of this section.
Increasing the market share is yet another reason for which a company engages in M&A activities. This is a facile way of gaining a higher market share considering the fact that
collaboration after an association between the firms was performed is perfectly legal, while colluding before such an action is against the law (Erel, Liao and Weisbach, 2004).
4 reduction (Brakman, Garretsen and van Marrewijk, 2006), quality assurance or economies of scale (Walker, 2000).
Lastly, diversification can guide firms towards M&As and it can be of two types.
Overlapping which presumes buying a different firm which can sell similar products or be active in the same market or industry and not overlapping diversification which implies the fact that the firm buys a completely unrelated firm, aiming to move to a better line of business which can be more profitable.
This approach may lead to obtaining less returns and it may take more time to achieve (Jensen, 1986). Diversification can also be split into other two types according to (Straub, 2007): internal and external growth, meaning that for internal growth developing your business
internally on the market on which the firm is already active in or getting hands on knowledge or capabilities of competitors through M&As is the strategy, while for external growth firms have to move to different countries or industries, a phenomenon which is known as FDI and which will be discussed more extensively in the subsequent paragraphs.
2.2. Foreign Direct Investment (FDI)
FDI is an important part of this research because it incorporates cross-border M&As, thus it is often the case that theories tailored for FDI are applicable to M&As (Brakman, Garretsen and van Marrewijk, 2006; Erel, Liao and Weisbach, 2004; Hyun and Kim, 2010).
After being mentioned for the first time by Coase (1937), the FDI started to get increased attention from more and more scholars and theories have been developed. Among those, perhaps the most famous one is the OLI theory which emerged when Dunning (1981) wanted to prove that the reasons behind a company’s decision to FDI can all be fitted into a framework. To that moment, the theories would only address isolated aspects relating to FDI, but Dunning managed to incorporate all those theories into one single model structured on the main three types of advantages a firm should hold in order to turn to FDI: Ownership (specific advantages which the investor exploits and which differentiate it from the competition), Location (advantages which the prospective country of investment has to offer to the company) and Internalization (choosing the type of entry - the higher the internalization benefits are, the lower the chances of licensing or franchising (Falk, 2016)).
5 categorization scheme which separates the reasons for FDI and not a proper theoretical model (Brakman, Garretsen and van Marrewijk, 2006).
Such a model was developed by Antras and Helpman (2004) and it can be used for studying the global outsourcing strategies. The choices a company was faced with were related to location, either stay in the domestic market or move to a foreign one, and organizational form. It has been shown that the prevalence of organization form depends on more factors such as: wage differences between countries, trading costs, productivity dispersion within a sector, bargaining power distribution, ownership advantage and intensity of headquartered services. All these factors could be regarded as FDI determinants.
There are other reasons which make FDI such a pervasive practice, among which is the positive implication it has on a larger category of aspects. Apart from being a development opportunity for the company engaging in it, FDI can also represent a growth factor for the host country economy, being regarded as a “good type of capital” (Daude and Fratzscher, 2007). This positive effect that FDI has is also proved by the increased competition among states to
implement policies which will provide incentives for foreign investors (Hyun and Kim, 2010). When diving more into the subject it can be seen that FDI is not just of one kind and it can differ according to the strategy of the firm. If a company is interested in making profit on the large markets and choose the host country based on the high wages of the people in the detriment of cost reduction, then it is more likely that this market seeking company will engage in what is called horizontal FDI. On the other hand, a firm which aims to reduce its production costs and seeks a host country based on the factor market motive will proceed with vertical FDI (Brakman, Garretsen and Marrewijk, 2006).
Another classification of FDI is related to the strategy of penetrating the market. Among others, Brakman, Garretsen and Marrewijk (2006) stated that FDI can be classified in two categories: Greenfield and M&As.
Greenfield is the type of investment that starts from nothing and implies that all the facilities required are built from scratch as a result of an organic growth. On the other hand, M&As are the situations when the firm would prefer to either acquire or merge with a company which already exists in the host country.
6 and loans from foreign parents as other types of foreign investment (Erel, Liao and Weisbach, 2012). Acknowledging the different ways of measuring the non-M&A component of FDI from country to country and the difficulty of developing a “one size fits all” type of measurement, this study is focusing only on cross-border M&As.
2.3. Cross-border M&As
M&As represent the largest part of the total foreign direct investment. A percentage of 78% of the total amount of worldwide FDI was consisting on M&As according to Brakman, Garretsen and Marrewijk (2010). This high percentage shows the crucial role which M&As play in the global investment setting.
However, the term of M&As used in this classification is referring to mergers and acquisitions conducted at an international level, namely, cross-border M&As. A definition for this concept was provided by UNCTAD (2000, p. 99), which sees cross-border M&As occurring when “control of assets and operations is transferred from a local to a foreign company”.
In broad lines, there is an agreement amongst scholars that the main reason behind cross-border M&As is the possibility of profit, this being the ultimate goal of every company.
Nonetheless, the literature shows that there are contradicting arguments regarding the underlying reasons of conducting such activities. On one hand, there is the argument that by being a sub-type of M&A, the reasons leading to a decision of cross-border M&As are the same as the reasons for domestic M&As, thus mainly related to value creation (Erel, Liao and Weisbach, 2012). On the other hand, there have been scholars which support the idea that cross-border M&As have the same motivations as FDI, since they are a type of foreign investment (Li, Li and Wang, 2016). Developing competitive advantage, diversifying risk and better exploitation of assets are among the mentioned motivators.
An argument for engaging in cross-border M&As is based on the internationalization theory and it assumes that having superior assets leads to the firm making the decision to exploit them on a different market which can prove beneficial in terms of achieving competitive
advantage in the new country, with the company being able to overcome drawbacks faced in the domestic institutional environment (Barkema and Drogendijk, 2007).
7 institutional restrictions, decrease in learning costs and cost savings gained from shared
production. All of those seem to be in line with the internationalization theory as they suggest that moving to another country would be cheaper than moving the product to that market by any other means.
A different argument supporting the fact that market imperfections act as an inducement for the bidder pushing it towards moving to a foreign country, has been presented by some papers (Conn and Connell, 1990), making the reasoning behind cross-border M&As a topic which is still debatable. In addition, Erel, Liao and Weisbach (2012) also consider market imperfections to be a driver of M&A and use changes in exchange rates along with stock market valuations in foreign currency as examples of imperfections which can be exploited for profit. Mostly this happens when overvalued firms are involved, because they are looking to purchase undervalued targets so that they can transfer value from the target to the bidder and keep the share price up. This is not due to the private information the management team of the bidder holds, but more likely because the target management is willing to accept payment in overvalued stock (Erel, Liao and Weisbach, 2012). An argument that is able to motivate the choice of
accepting overvalued stock as payment is given by John el al. (2010), who makes a point that stock payment can create trust with both the bidder and target shareholders sharing the risks.
Seeing market imperfections as a motivator of cross-border M&As is feeding the idea that the majority of cross-border deals occur between a bidder from a well-developed economy and a target from an emerging market (Daude and Fratzscher, 2007). The so called “Fire-Sale FDI” which presumes that firms from countries in crisis are sold to bidders from developed economies at prices lower than the actual value (Erel, Liao and Weisbach, 2012), is in line with this idea of deals happening mostly between developed and emerging economies. However, the Lucas paradox contradicts this reasoning and shows that 70% of the global capital flows is between developed countries (Hyun and Kim, 2010).
Overall, this type of foreign investment seems to have only positive implications as it supports growth in the host countries by transferring technology and knowledge or by opening the market abroad (Daude and Fratzscher, 2007).
8 influenced by the differences (especially cultural) between the home and host country (Basuil and Data, 2015). Uncertainty and information asymmetry are caused by the “liability of foreignness” which is a specific term used to describe the cumulating factors which make it harder for a foreign firm to perform in the new market. Opposing this thinking is the argument of Daude and Fratzscher (2007) who believe that FDI and implicitly cross-border M&As, is able to circumvent information asymmetry completely.
Usually those differences go under the name of “distance”, which is a specific term in the academic literature concerning internationalization. Many articles have been written about it and it is considered that distance is one of the most influencing factors when it comes to an
investment being successful. Although it is agreed that distance between the two countries matters and can eventually affect the performance of the investment, there is a lack of agreement regarding which type of distance is important as researchers took turns in focusing on different types of distances: cultural distance (Hofstede, 2001), geographic distance or institutional distance all applicable to their researched topics. There have been attempts to standardize the concept of distance between countries by defining a contextual distance which is a
multidimensional type of distance, incorporating: economic, language, religious, cultural and geographic distance (De Jong et al., 2015). The general impression is that distance has a negative effect on the success of an investment. Therefore shorter distance between countries is associated with an increased likelihood of acquisitions (Erel, Liao and Weisbach, 2012). Although this is generally accepted, the FDI theory proposes both a positive and a negative effect of distance depending on the reasons behind the investment (Hyun and Kim, 2010).
9 2.4. Corruption
One of the most important components of the institutional setting from a country is corruption. Even though corruption itself is not an institution, the integrity and interaction of institutions either allow for corruption to be present or hamper it. The importance of this phenomenon is given by its worldwide spread and the incredibly high amount of prejudice it is creating with approximately 1 trillion USD quantified as annual briberies (Kaufmann, 2005). Corruption itself is a vague term, being defined in multiple ways along the years. One definition which became popular is the one of Rodriguez et al. (2005) which presents corruption as being “the abuse of public power in order to obtain a private benefit”. This is built on a definition included in what is perhaps one of the breakthrough studies on corruption which presented corruption as “the sale by government officials of public property for personal gain” (Shleifer and Vishny, 1993).
Many studies decided to focus on corruption but mostly, the link between corruption and FDI was studied with the cross-border M&As being overlooked (Weitzel and Berns, 2006). Fewer studies (Henisz, 2000; Smarzynska and Wei, 2000) look at the effect corruption has over cross-border M&As, taking into account country, firm and deal characteristics. This is a gap which needs to be addressed as the cross-border M&As represent the most significant form of FDI, totaling, in 2006, to approximately 72% of the entire FDI volume (Malhotra et al., 2010). It can be called a paradox, the fact that even with the extensive literature on corruption, a solid theoretical framework linking corruption with M&As is still to be developed. This is due to most of the studies being a-theoretical (Judge, et al., 2011).
A model for the phenomenon of corruption has been created by Shleifer and Vishny (1993) but it discusses corruption as a general matter without focusing on its effects over a certain field. As many other studies, this one looks at corruption through a principal – agent perspective and is mostly focused on bribes, overlooking other forms of corruption.
Some studies turn to theory and try to explain corruption as a combination between the institutional view and game theory (Judge, et al., 2011), which could be valid considering the fact that most of the time corruption is associated with the public sector and is thought of being largely present in countries with a weak institutional environment (Malhotra et al., 2010). To further develop the institutional perspective on corruption, Li et al. (2007) mention that
10 institutional setting will provide the chance for companies to engage in illegal actions. From the perspective of game theory, corruption is a concept which includes individual actors, each having their own interest in a competitive environment such as the market. The choices they are faced with are to cooperate and provide advantages for one another, do things the right way and ensure loyal competition on the market or one of the parties involved turns to corruption facing the risk of being turned down and even denounced. Game theory represents the base of other corruption related reasoning such as the lock in effect which suggests that parties involved in a corruption act are tied together even after the deed is done as the risk of denunciation by any of the parties is still present (Weitzel and Berns, 2006).
There are two different views in the literature: one which sees corruption as being value destroying by increasing costs and risks (Egger and Winner, 2005; Judge et. al, 2011; Cuervo-Cazurra, 2006) and a second one which launches the idea that corruption can also be useful and has a positive effect on business by easing laborious processes (Judge et. al, 2011; De Jong et. al, 2010; Egger and Winner, 2005; Kaufmann and Wei, 2000). These views received representative names being called “the grabbing hand” and “the helping hand”.
11 which require more time, procedures and involve more people, creating more possibilities for bribery (Stansbury, 2005).
On the other side there is the corruption acting as a “helping hand”. This implies the fact that such a negative practice can actually have positive implications over business. One argument is that firms which come from a corrupt country and are used to integrating corruption in their activities and doing business this way, may benefit from investing in another corrupt market as they are already familiar with the processes (Habib and Zurawicki, 2002). Another argument is given by Egger and Winner (2005) who say that in some cases corruption is seen as a stimulus for investment. A valid point is that in a corrupt environment the excessive bureaucracy also known as red tape is slowing processes and that can be avoided also by forms of corruption such as bribes (Cuervo-Cazurra, 2006). The speeding up of those bureaucratic processes is referred to as “grease money” (De Jong et. al, 2010). The illegal characteristic of corruption can also be helpful as it can lead to increased trust between the involved parties, considering the fact they are in it together and all are breaking the law. Developing this idea, corruption can also lead to building an improved network as the people trusting you, can introduce you to other different people thus reducing the “liability of foreigness” (De Jong et. al, 2010). However, this
hypothesis of “efficient grease” is limited when considering the tendency of the officials to build wealth by exploiting this red tape opportunity and charging multiple fees per service or charging in accordance with the client’s disponibility of payment (Kaufmann and Wei, 2000).
Taking all the contradicting views into account, corruption still remains a set of practices which is associated with stepping out of the rule of law, being tantamount to deregulation (Kaufmann and Wei, 2010). The mostly spread perception is that corruption hardens the fair and efficient wealth distribution at a worldwide level (Judge et. al, 2011). Considering all this and the common sense interpretation of corruption, it is believed that it would more likely harm value creation rather than enhance it. In addition, Kaufmann and Wei (2000) did not find evidence to support the “grease money” argument, therefore, the following hypothesis is developed:
12 2.5. Corporate governance
To stress out the importance of the institutional quality in the context of cross-border M&As even more, corporate governance is considered when assessing the effect of corruption because it is extremely important to analyze other variables related to the concept of governance as they can explain similar effects (Weitzel and Berns, 2006). Consequently, corporate
governance is the firm level moderator included in this study.
Corporate governance is an aspect of business which started to attract a lot of interest in the past decades. It developed, increased in depth and became more and more complex with the years passing by, but one of the most important events which led to this is LaPorta’s paper (1998) which established a link between the quality of legal protection and financial
development. From that point on, theories have been developed and corporate governance was analyzed at both country and firm level.
Considering the fact that there are codes of good corporate governance and that everyone talks about it in general terms, it is hard to put a finger on what corporate governance exactly is. Definitions have been proposed throughout time but being concise about a term so broad is a difficult task. One of the definitions which comes close to the truth and at the same time is concise, was proposed by Tirole (2001) and defines corporate governance as being “the design
of institutions that induce or force management to internalize the welfare of the stakeholders”.
An interpretation of this quote leads to believing that corporate governance is about having the management act in a way that will satisfy all the company’s stakeholders. However, corporate governance includes some other dimensions as it is stated by Blair (1995). It is also about what public companies can do, the way in which control over them is exercised, who is in control and who takes on the risk and returns with which the companies are faced.
When linking corporate governance with value creation, it can be seen that multiple studies have been conducted in order to see in which way corporate governance affects the firm’s value (Danbolt and Maciver, 2012; Masulis, Wang and Xie, 2007 or Starks and Wei, 2013).
For understanding the entire process, first, the way in which corporate governance changes in the situation of M&As needs to be clarified. In case of a full-takeover the corporate governance of the target is replaced (Martynova and Renneboog, 2008). Starting from here, more hypotheses were developed in regards to the effect of corporate governance changes on
13 better corporate governance than the target and by replacing the latter, the firm gets a benefit. This ultimately translates into the company recording higher returns. This hypothesis has been tested by Martynova and Rennenoog (2008) and was found to be valid. The same study proposes a different hypothesis which is the antithesis of the previous. It is believed that a worse bidder corporate governance will mean negative returns and value destruction. However, no evidence supporting this negative spillover effect were found, therefore it remains uncertain.
One motivation for the negative spillover not being valid is proposed by Stark and Wei (2013) who make a case that, when firms are involved into a merger or an acquisition, the better corporate governance system will prevail. On the same note, there are voices stating that
corporate governance can be portable and, if the acquired firm has a lower level of corporate governance, it can adopt the one of the acquirer (Ellis et al., 2016). This means that it is not necessarily a condition that the bidder imposes its system on the target, but there is also the possibility of it adopting the target’s governance system or incorporate parts of it which are thought to improve the efficiency of the system.
In spite of all that, the idea that bad corporate governance can decrease the bidder firm value is supported by evidence. There is however, a shortcoming of the paper researching this topic (Giroud and Muller, 2011), which is related to its generalizability of the findings, due to the fact that it focuses on noncompetitive industries.
Attention has been also given to partial takeovers and the results show that the process is not entirely different from the full-takeover, with the bidder being capable of imposing its corporate governance system on the target as it gained control over it (Martynova and Rennenoog, 2008).
Building on the argument that a good corporate governance of the bidder will translate into positive returns, there are some studies which support this idea providing different
14 Shareholders relationship (Masulis, Wang and Xie, 2007). Furthermore, Danbolt and Maciver (2012) suggest that a strong corporate governance code of the bidder may be able to stop the managers from undertaking acquisitions which destroy value. A study conducted by Kuipers et al. (2009) proves the fact that bidders with higher corporate governance end up paying smaller amounts in terms of premiums for the desired targets. The meaning of this is that better corporate governance creates the possibility of having positive returns.
Based on the theoretical background which has been reviewed in this section and on the argument that a solid corporate governance of the bidder can prevent the managers from
engaging in value destroying acquisitions, the following hypothesis is tested:
15
3. Data and Methodology
This section will firstly discuss the approach for gathering and compiling the data and will continue with a more in depth description of the methodology used for this research which consists in a description of the variables and their measurements. Towards the end, the regression models will be presented, one for each of the formulated hypothesis.
3.1. Sample
In order to investigate the relationship between corruption of the target country and the bidder’s shareholders’ wealth, a sample of M&As was created. The Zephyr database was used to identify the deals which fit the requirements of this study.
First of all, only the deals which involved a bidder which was publicly listed were taken into account. The main reason for choosing this filter had to do with data availability. The fact that more data is available for a public company than for a private one is intuitive but at the same time has been empirically proven.
A second filter was included so that the search would only return M&As and no other types of deals. Another condition was that the deals are cross-border, meaning that the bidder and the target come from two different countries. These last two conditions were motivated by the fact that cross-border mergers and acquisitions represented no less than 72% of the total amount of global FDI (Malhotra et al., 2010). As this paper calculates abnormal returns for a period of 3 years the time frame chosen was 2007-2014 due to lack of data for more recent deals.
The method of payment is an important aspect to be considered in the M&As research field, this being proved by multiple studies that show the different types of influences it can have on a deal (John et al., 2010; Martynova and Renneboog, 2008; Masulis, Wang and Xie, 2007). A filter was set for including only the mergers and acquisitions for which the payment was made in cash or in shares.
16 One of the most important filters is related to the bidder origin. Since the availability of data is a real issue, and the bidder is the central point of this research, the countries of origin were chosen to be the USA and the UK because of their common law based legal system which encourages companies to disclose more information.
Lastly, the deals included in the sample had to have some importance to the market. Therefore the deal value was chosen to be greater than 10 million USD.
Finally, the sample counted a total 193 deals each with fully available data.
3.2. Variable description
This part is going to focus on the variables included in this study and the methodology used for measuring them, as well as reporting the sources used to gather the data.
3.2.1. Dependent variable
The dependent variable is the shareholders’ wealth. This has been the subject of many previous investigations (Doukas and Travlos, 1988; John et al., 2010; Barber and Lyon, 1997). The majority of the studies so far use the concept of abnormal returns as a proxy for
shareholders’ wealth. However, it has been highly debated which way of computing the abnormal returns is the most comprehensive.
Two main reasoning trends have been put head to head and both propose a different angle in seeing this matter of abnormal returns. The first one is the calculation of short term abnormal returns which is done around the announcement date of the deal. This approach uses a rather short time frame, focusing on periods starting from one day and going up to a few months the latest. It is a highly popular method and it has been used in a large number of studies by the likes of Brown and Warner (1980), Dyckman et al. (1984), Doukas and Travlos (1988) and more recently by John et al. (2010).
17 BHAR are calculated in a different way than the CAR, with the main difference being that instead of having a sum all the abnormal returns obtained within the selected period, BHAR uses the product of all the daily or monthly returns from which it subtracts the product of the expected return. It has become the case that instead of the expected return, the scholars rely on the market return for that specific period and that market return is either obtained by creating a portfolio made out of the best performing stocks or by using an already existent index (Barber and Lyon, 1996). Although there has been some criticism recorded in relation to this approach, and other methods of computing the BHAR have been developed, this study is going to make use of the market return as a proxy for the expected return due to issues regarding data availability and the precedent set by the large number of scholars using the same method in their studies (Barber and Lyon, 1996).
The corresponding formula used in this study for the calculation of the long run abnormal returns is:
𝐵𝐻𝐴𝑅𝑖𝑡 = ∏𝑡𝑖=1(1 +𝑅𝑖𝑡) − ∏𝑡𝑖=1(1 +𝑅𝑚𝑡) (1) Where BHAR means the buy and hold abnormal returns, i is related to the firm selected from the sample, t represents the time period – usually the number of the month for which the return is calculated, R is the return and m is the parameter of the market return.
As previously mentioned, this approach may include some limitations and biases such as the new listing bias. The new listing bias is related to the risk of the portfolio chosen as reference may include newly listed firms while the sample only has firms with trading history. The
problem with the newly listed firms is that those might be underperforming because of their lack of experience on the market (Ritter, 1991). However, by choosing a return on an equally
weighted market index, the risk of encountering this bias is substantially reduced as it is difficult for newly public firms to be selected in one of those indexes.
18 Regarding the timeframe, a period of three years was chosen for investigation because only looking at announcement returns may lead to drawing false conclusions as the value fluctuations can be caused by other factors than the ones included in the scope of the research.
Considering the fact that this paper focuses on the long term approach, the BHAR
method is more suited as it allows the abnormal returns to be observed over a longer time period and not just the announcement window (Brown and Warner, 1985). Moreover, there are other limitations to using CAR over longer periods, such as the risk of short windows not being able to grasp the economic impact of a complex deal or misinterpret the implications of the
announcement returns and make predictions about the actual performance (Oler et al., 2008). In terms of data, it was collected using DATASTREAM after matching the acquirer with the corresponding returns obtained for a period of 36 months after the announcement date of the deal. This matching between the two utilized databases, Zephyr and DATASTREAM, was performed using the ISIN number of the acquirer. The RI symbol from DATASTREAM
corresponds to the return of the company but the fact that it is calculated by adding or subtracting the obtained result to the value from the previous day means that in most of the cases the value will be extremely high and there would be no common ground which will allow for a comparison among firms. For addressing this matter, the change from one month to the other was computed as a percentage, creating thus a similarity in the values of the returns for all the different firms included in the sample.
All the acquirers are listed companies in the USA or the UK, consequently when
analyzing the market return the two most important indexes were considered respectively: S&P 500 for the US and FTSE 100 for the UK. The same rationale for computing the market returns was used as for the firm return.
Finally, the obtained values were plugged in equation (1) and the BHAR were computed.
3.2.2. Independent variable
19 The score ranges from 0 which is associated to a perfectly clean country, to 100 which translates to a highly corrupt public sector (Transparency International, 2017). The time when the deal took place is accounted for and the value of corruption from the target country is collected from the report issued in the same year as the deal happened. The score measurement has recently been changed moving from a 0 to 10 range to a 0 to 100. To be able to have comparable scores, the old ones were multiplied by 10, managing to fit all in the same measurement system.
Another popular way of quantifying corruption is the Control of Corruption Index provided by the World Bank. It measures how well does a country control its corruption and it varies between -2.5 and 2.5 with higher values meaning a better control of corruption (Judge, McNatt and Xu, 2011). This index is used when performing the robustness test.
3.2.3. Moderator
The selected moderator for this relationship is corporate governance of the bidding firm. Most of the studies that are dealing with corporate governance treat it at a country level (Starks and Wei, 2013; Martynova and Renneboog, 2008), but this paper is considering corporate governance at a firm level due to several reasons. The fact that the acquirers are originally from two different countries, would eliminate any diversity in the sample as the corporate governance scores will only be varying depending on the year of the deal. The score has been obtained from the ASSET4 database by matching the sample with the existing firm observations of the database using the ISIN number of the acquirer. In extracting the data the indicator used for corporate governance was CGVSCORE, which reflects the extent to which a company has adopted processes and systems that can assure the shareholders that both the management and the board members act in the sole long term interest of the share-owners. In addition it also shows how capable the company is to create shareholder value by using management best practices (ASSET4 ESG DATA GLOSSARY, 2013). The score ranges from 0 to 100 and has a positive implication meaning that 0 is the worst score while 100 implies a perfect level of corporate governance.
3.2.4. Control variables
20 This requires further analysis in order to make sure that the results are not biased by those
factors. Therefore a set of control variables which is specific to the M&A literature has been accounted for. These controls have been separated in three different groups: bidder
characteristics, deal characteristics and country characteristics.
Bidder characteristics
Bidder size is controlled for and it is going to be documented as firm size. The main reason behind this has to do with the hubris hypothesis developed by Roll (1986) which stated that the more confidence a firm or management team has, the more likely it is for it to overlook aspects and make mistakes due to an excess of self-confidence. In accordance, the bigger the acquirer is, the bigger the chance of overpaying for a target gets. This could eventually translate in the possibility of having lower returns, creating a negative correlation between firm size and the acquirer returns (Martynova and Renneboog, 2008). This variable is calculated as the natural logarithm of the bidder’s total assets (Garfinkel and Hankins, 2011).
Another bidder characteristic to be controlled for is the leverage. The argument for this control resides in Jensen’s free cash flow hypothesis (1986) which states that a low level of leverage in combination with high cash flows increases the probability of management engaging in value destroying acquisitions which are going to create negative returns (Martynova and Renneboog, 2008). This together with the fact that managers are under creditor’s pressure suggests that leverage may have a positive effect the shareholders’ returns.
According to Basuil and Datta (2015), pre-deal firm profitability should also be
controlled for as this may affect the returns gained post-deal. A firm which was profitable before, can still benefit from it and mislead into believing that the positive returns are due to the newly acquired firm. Profitability was calculated as the average of ROE recorded in the past three years before the M&A happened.
21 Deal characteristics
In terms of deal characteristics, the first thing which should be taken into consideration is the method of payment, as many M&As related studies decide to focus on it (John et al., 2010; Martynova and Renneboog, 2008; Masulis, Wang and Xie, 2007). However, evidence regarding which method of payment is generating better returns are mixed. Masulis, Wang and Xie (2007) reported that equity payment generates significantly negative returns while, on the other hand, John et al. (2010) suggest that paying by stock may lead to increased trust as both the bidder and target shareholders become share owners in the new company and share the risks. This may signal the fact that the stock is not overvalued and positive returns may be recorded. A dummy variable is created to account for the method of payment. A value of 1 was attributed for every deal where the payment was made by equity and all cash deals received a value of 0.
Another variable which is controlled for is the similarity of the industry. There is a spread belief across the literature that acquisitions made within the same industry (horizontal
acquisitions) are increasing shareholder value (Martynova and Renneboog, 2008; Oler et al., 2008). Therefore, after grouping the transactions based on the first two digits of the firms SIC codes, a dummy variable was generated to take a value of “1” if the industry of the target belonged to the same group as the industry of the bidder and “0” otherwise.
Lastly, the target size is considered to be the logarithm of deal value (John et al.,2010). Returns increase with the target size meaning that between these two variables there is a positive relationship (Masulis, Wang and Xie, 2007). Conceptually this is valid as the more a firm pays the more it expects to get out of the deal, but there may be cases when other interests may influence the deal and returns are therefore negative.
Country characteristics
22 3.3. Model
In order to test for the two hypothesis, two different models are going to be used. First, the relationship between target country corruption and shareholders’ value of the bidding firm is considered and the following model is utilized:
𝐵𝐻𝐴𝑅 = 𝛼 + 𝛽1∗ 𝐶𝑜𝑟𝑟𝑢𝑝𝑡𝑖𝑜𝑛 + 𝛽2∗ 𝐹𝑆 + 𝛽3∗ 𝐿𝑣𝑔 + 𝛽4 ∗ 𝑅𝑂𝐸 + 𝛽5∗ 𝐿𝑖𝑞 + 𝛽6∗ 𝐴𝐶𝐶 + 𝛽7∗ 𝑀𝑜𝑃 + 𝛽8∗ 𝑇𝑔𝑆 + 𝛽9∗ 𝑆𝐼 + 𝜇 (2)
Where BHAR stands for buy and hold abnormal returns, FS represents bidder size, Lvg is the leverage of the bidder, ROE represents the three year average of the bidder return on equity, Liq is the liquidity of the acquirer, ACC are the accounting standards from the target country, MoP is the dummy generated for the payment method, TgS represents the size of the target, SI stands for industry similarity and 𝜇 is the error term.
The second model is helpful when analyzing the impact of the bidder’s corporate
governance over the relationship between corruption and shareholder value. The final form of the model is:
𝐹𝑉 = 𝛼 + 𝛽1∗ 𝐶𝑜𝑟𝑟𝑢𝑝𝑡𝑖𝑜𝑛 + 𝛽2∗ 𝐶𝐺 + 𝛽3∗ (𝐶𝑜𝑟𝑟𝑢𝑝𝑡𝑖𝑜𝑛 ∗ 𝐶𝐺) + 𝛽4∗ 𝐹𝑆 + 𝛽6∗ 𝐿𝑣𝑔 + 𝛽7∗
𝑅𝑂𝐸 + 𝛽8∗ 𝐿𝑖𝑞 + 𝛽9∗ 𝐴𝐶𝐶 + 𝛽10∗ 𝑀𝑜𝑃 + 𝛽11∗ 𝑇𝑔𝑆 + 𝛽12∗ 𝑆𝐼 + 𝜇 (3)
WhereCG stands for the corporate governance score of the bidder.
3.4. Sample distribution
The distribution of the sample is presented and is mostly focused on analyzing the years and countries where the deals happened as well as the industries which the firms belong to.
23 effects of the crisis were only felt in the following financial year, thus the decrease occurred in 2009. However, there is another significant increase for 2010 which is surprising even if the fact that 2010 meant the beginning of the end when talking about the financial crisis. The number of M&As, which is more than double when compared to the previous year cannot be entirely explained by the crisis, being close to the end, but much rather by a combination of this argument and the one presented earlier concerning the public nature and size of the firms included in the sample.
Table 1: Yearly distribution of cross-border M&As
Acquirer Country Year UK US Total 2007 12 6 18 2008 12 13 25 2009 7 8 15 2010 20 18 38 2011 21 18 39 2012 9 16 25 2013 10 17 27 2014 0 6 6 Total 91 102 193
The table shows the entire sample of cross-border M&A which took place in the eight year time frame separated by the year in which the deals happened. This is reflected by the last column which accounts the total number of deals per year. The second and third column split the number of transactions based on the country of origin of the acquirer.
Next, it can be seen (Appendix 2) that the sample of 193 deals is divided between 25 countries from the whole world, ensuring this way the sample diversity.
Despite the large number of target countries included in this analysis, it can easily be seen that most of the deals are concentrated in the United States and the United Kingdom. The
24 Lastly, Table 2 gives an overview of the industries from which the companies involved in all the deals come from. When grouping the industries into clusters, the classification was based on the US standard industrialization code (SIC code). Division F – Wholesale trade along with division J – Public administration were excluded from the table as the sample did not contain any firms belonging to those industries. The majority of the companies come from either Construction or Services industries, which is understandable considering the fact that one of the main reasons driving the cross-border deals is cost reduction which can be achieved by internalizing production costs and reducing transportation costs (Construction industry) or by reducing distance between provider and client (Services industry). This trend is similar for both the acquirer and target with the Manufacturing industry giving 45.18% of the total number of the targets and 56.85% of the acquirers. The Services industry has a lower percentage of both targets (35.53%) and acquirers (25.89%) but is still by far the second best represented industry from this sample.
Table 2: Distribution of cross-border M&As across industries
SIC Code Target industry % Acquirer industry %
A 01 - 09 Agriculture, forestry and fishing 5 2.59 5 2.59
B 10 - 14 Mining 3 1.55 0 0.00
C 15 - 17 Construction 87 45.08 110 56.99
D 20 - 39 Manufacturing 12 6.22 11 5.70
E 40 - 49 Transportation, Communications 8 4.15 5 2.59
G 52 - 59 Retail trade 2 1.04 7 3.63
H 60 - 67 Finance Insurance and Real Estate 8 4.15 6 3.11
I 70 - 89 Services 68 35.23 49 25.39
25
4. Empirical results
In this section a short analysis of the dataset is performed in order to guarantee
conformity of the variables when the regression is tested. Firstly, descriptive statistics are shown in Table 4. Subsequently the outcome of the regression analysis are presented and used for testing the hypothesis formulated in the first part of the paper.
4.1. Descriptive statistics
The positive mean value of BHAR shows that overall, firms from this sample record positive returns after a period of three years from the announcement date. Furthermore, the corruption levels of the target countries seems to be low as it is suggested by the high values of the mean for the two different variables measuring corruption. This is due to the fact that most of the deals went through in either US or UK, countries known for being stable markets and having a reduced level of corruption.
26 Table 3: Descriptive statistics
Variable Obs Mean Std. Dev. Min Max
BHAR 193 0. 2584069 0.8867296 -1.276361 7.619939 CorruptionTI 193 75. 62694 12.08372 30 94 CorruptionWB 193 1. 549692 0.5817414 -0.603604 2.530376 CGV 193 67. 03591 26.63104 1 95.17 CTIxCGV 193 5037.6 2174.999 71 8365.35 CWBxCGV 193 102.0904 58.72735 -57.1613 216.213 Biddersize 193 14.2488 1.955226 7.201171 20.47713 Leverage 193 20.9514 19.16539 0 167.24 ROE 193 -2.348178 220.3856 -2103.77 455.2833 Quickratio 193 1.446218 1.294561 0.03 9.82 Targetsize 193 11.40983 1.465731 9.259131 16.20427 Accountingstandards 193 70.31606 7.249181 24 83
The table displays the descriptive statistics of the most important variables which are used in the regression analysis. BHAR constitutes the dependent variable, while CorruptionTI is the main independent variable. This is followed by CGV which is an independent variable and the moderator CTIxCGV. The control variables are specific to the field of M&A scientific research. An important note is that dummy variables were left out of the descriptives as their values do not shed any light in this particular situation not playing a major role in the development of hypothesis. The variables used for robustness check were also included as being CorruptionWB and CWBxCGV.
4.2. Multicollinearity
Multicollinearity among variables can often be an issue which can affect the standard errors of the estimated parameters. To test for multicollinearity, Table 4 presents a correlation matrix of the most important variables included in the regression analysis. As it can be observed from the table there are no cases of severe correlation between the variables except for a few exceptions. There is a high correlation between the two variables used for measuring corruption but this does not represent an issue as CorruptionWB is used in the robustness check to substitute CorruptionTI.
27
Table 4: Multicollinearity Matrix
The table shows the level of correlation between the most important variables included in the regression analysis. The main independent variables are
CorruptionTI and CGV, with CTIxCGV being the moderator. The dependent variable is BHAR and the rest of the variables included in the table represent the controls. CorruptionWB and CWBxCGV are used for the robustness check.
BHAR Corruption TI Corruption WB CGV CTIx CGV CWBx CGV Bidder
size Leverage ROE
28 4.3. Regression results
This section presents the outcome of the regression analysis and uses the results for testing the hypotheses formulated in the first part of the paper. To achieve this purpose, linear regression was used, all regressions being performed in Stata with robust errors taken into account.
Table 5 provides an overview of all the regression outputs obtained when the first hypothesis was tested and shows the results of a test made in relation to the relevance of the chosen controls.
29 horizontal acquisitions seems to have a negative impact on shareholder value, contradicting arguments found in the literature (Martynova and Renneboog, 2006).
For testing the first hypothesis, the effect corruption has on the long run abnormal returns, five different models were created, all having the BHAR as the response variable and corruption as explanatory variable. Therefore, the first model looks strictly at the link between these two and confirms the expectation that there is a negative relationship between them. The reliability of this model is however low considering the fact that the reported value of the R-squared is 0.011, meaning that this model can only predict 1.1% of the variation in returns. F-statistic is significant at a 5% level, meaning that the null hypothesis of there being no effect is rejected. This translates into an impossibility to base conclusions on it. Moreover, corruption, the main explanatory variable, is not significant although the coefficient indicate a negative
relationship which would be supporting the hypothesis of corruption destroying value.
The second model included the control variables specific to the M&A research sphere. This helped as the model had an increased R-squared, and the F-statistic remained significant at 5% level. This model is capable of explaining 8.2% of the variation in BHAR, which is still a rather small percentage. Method of payment also proves to be significant at a 1% level and strengthens the evidence found by Masulis et al. (2007) that payment in shares negatively affects the returns. This could be caused by the fact that a cash offer shows the bidder has more
confidence in the target as it tries to buy out the shareholders and keep the future returns for itself. Judging by the p-value of the corruption variable, there is not enough evidence to support the hypothesis that corruption in the target country influences the abnormal returns, therefore the null hypothesis is not rejected in this case as corruption has no effect over the abnormal returns. Overall, the model shows that it has effects on the abnormal returns but the main variable of interest, corruption, has not. Consequently hypothesis one cannot be confirmed.
30 Significance of the variables is mostly unchanged, the same variables as in model 2 remaining important. One change is recorded by the bidder size which becomes significant in model 3 and 5, helping thus to confirm the hubris hypothesis. An interesting result is the fact that accounting standards coefficient is changing sign for the third model, becoming negative. This is contradicting the literature (John et al., 2010) and the common sense assumption that making a more informed decision is going to generate increased returns. When comparing the R-squared of the three models it can be seen that there is a constant increase from model to model, reaching 0.179 in the last one where both year and industry fixed effects are accounted for. This is still a fairly low value that does not guarantee a comprehensive model. Without any improvement in the significance level of corruption, the first hypothesis cannot be backed.
Overall, all models seem to point towards a negative relationship between corruption in the target country and the bidder returns. A decrease of the perceived corruption by 1 unit (worse corruption) is expected to generate 0.7% to 0.9% lower buy and hold abnormal returns,
depending on the chosen model. The most comprehensive one (the one with the highest R-squared) suggests returns lower by 0.7%.
31 Table 5: Regression results – Hypothesis 1
Hypothesis 1
VARIABLES Controls Model 1 Model 2 Model 3 Model 4 Model 5
CorruptionTI -0.0077 -0.0090 -0.0088 -0.0070 -0.0070 (0.0102) (0.0104) (0.0113) (0.0116) (0.0124) Biddersize -0.0521 -0.0535 -0.0763* -0.0597 -0.0825* (0.0431) (0.0437) (0.0460) (0.0463) (0.0481) Leverage 0.0020 0.00232 0.0020 0.00159 0.0015 (0.0034) (0.0033) (0.0028) (0.0036) (0.0036) ROE -0.0003 -0.0003 -0.0002 -0.0003 -0.0002 (0.0003) (0.0002) (0.0003) (0.0003) (0.0003) Quickratio -0.118* -0.119* -0.125** -0.131** -0.131** (0.0641) (0.0640) (0.0631) (0.0658) (0.0653) Targetsize -0.0760 -0.0743 -0.0489 -0.0710 -0.0529 (0.0477) (0.0471) (0.0469) (0.0493) (0.0513) AccS -0.0047 0.0024 -0.0006 0.0041 0.0018 (0.0099) (0.0082) (0.0086) (0.0083) (0.0085) MoP -0.826*** -0.884*** -1.007*** -0.995*** -1.101*** (0.298) (0.323) (0.288) (0.362) (0.326) SI -0.0915 -0.0974 -0.0878 -0.0641 -0.0628 (0.130) (0.133) (0.137) (0.141) (0.143) Constant 2.393* 0.837 2.574* 3.116** 2.753* 2.130 (1.287) (0.805) (1.431) (1.478) (1.539) (1.698)
Year fixed effects No No No Yes No Yes
Industry fixed effects No No No No Yes Yes
Observations 193 193 193 193 193 193
R-squared 0.070 0.011 0.082 0.108 0.157 0.179
F statistic 2.26** 0.57** 2.06** 2.21*** 9.54*** 3.22***
The following notation was used: *, **, *** for 10%, 5% and 1% significance levels.
32 Testing the second hypothesis, the same reasoning was applied and five different models were tested in order to verify the validity of the hypothesis that the corporate governance of the bidder can help in creating better returns when investing in a corrupt country.
The first model represents a regression which only includes the corruption level, the corporate governance score of the acquirer and the interaction variable. With the F-statistic being significant at a 1% level, the model can be used to analyze the variation in BHAR but it can only explain 5.4% of the variance of the returns. If the negative relationship between corruption and returns was expected, the indirect relation between returns and corporate governance is
surprising. It comes in contradiction with previous studies such as the ones conducted by Danbolt and Maciver (2012). However, the results show that when a company with high corporate governance invests in a highly corrupted environment, it can still expect positive returns. This is in line with the underlying assumption on which the second hypothesis was based, but it cannot be used to confirm it, as the moderator is not significant. Consequently, the hypothesis of the moderator having no effect on the returns is not rejected in case of the first model.
Further, controls are added in order to form the second model. This is a more
comprehensive model, fact shown by the increased p-value of the F-statistic and the increased R-squared (0.082). In terms of results, only controls prove to be significant and a negative relation between bidder size and the returns is supported at a 10% significance level, again providing evidence for Roll’s hubris hypothesis (1986) and being in line with the results of Martynova and Renneboog, 2006). Method of payment and quick ratio are relevant at a 95% respectively 90% confidence level, both manifesting negative coefficients and thus an indirect relationship with the bidder returns. Even though these variables are significant, neither the corporate governance nor the interaction variable are, making it impossible to support the hypothesis.
33 An interesting finding is that accounting standards are having a negative influence over the abnormal returns in each of the five models, being in complete opposition with previous literature (John et al., 2010). A reason for this could be the fact that accounting standards force companies to disclose more information, making it easier for potential acquirers to decide whether to buy the firm or not and reducing thus the possibilities of discovering “hidden gems”. However, the impact is not extremely harmful for the bidder as a decrease of 1 point in terms of accounting standards causes returns lower by 0.33% up to 0.81% depending on the regression model.
34 Table 6: Regression results – Hypothesis 2
The following notation was used: *, **, *** for 10%, 5% and 1% significance levels.
The table presents the results of the regressions including both the independent and the control variables. Each variable’s coefficient is reported on the same row and the variable and the values between brackets represent the standard error. MoP and SI are dummy variables generated based on the information obtained about the method of payment and industry similarity.
Hypothesis 2
VARIABLES Model 6 Model 7 Model 8 Model 9 Model 10
CorruptionTI -0.0522 -0.0504 -0.0492 -0.0497 -0.0480 (0.0440) (0.0442) (0.0458) (0.0470) (0.0482) CGV -0.0476 -0.0453 -0.0439 -0.0479 -0.0459 (0.0410) (0.0420) (0.0431) (0.0444) (0.0452) CTIxCGV 0.0007 0.0007 0.0006 0.0007 0.0006 (0.0005) (0.0005) (0.0005) (0.0006) (0.0006) Biddersize -0.0722* -0.0976** -0.0754 -0.0994** (0.0426) (0.0431) (0.0474) (0.0485) Leverage 0.0018 0.0016 0.0010 0.0010 (0.0033) (0.0029) (0.0038) (0.0037) ROE -0.0004 -0.0002 -0.0003 -0.0002 (0.0003) (0.0002) (0.0003) (0.0003) Quickratio -0.120* -0.125* -0.136* -0.135* (0.0672) (0.0658) (0.0703) (0.0694) Targetsize -0.0613 -0.0395 -0.0578 -0.0459 (0.0452) (0.0470) (0.0456) (0.0505) AccS -0.0049 -0.0081 -0.0033 -0.0057 (0.0097) (0.0099) (0.0101) (0.0098) MoP -0.770** -0.905*** -0.891** -0.997** (0.348) (0.331) (0.406) (0.388) SI -0.0716 -0.0607 -0.0344 -0.0310 (0.121) (0.124) (0.125) (0.127) Constant 4.180 6.209 6.726 6.428 5.708 (3.551) (4.183) (4.276) (4.381) (4.784)
Year fixed effects No No Yes No Yes
Industry fixed effects No No No Yes Yes
Observations 193 193 193 193 193
R-squared 0.054 0.126 0.152 0.195 0.215
35 4.1. Robustness test
In order to be assured about the reliability of the models used and of the obtained results, a robustness test was performed. This test focused only on the main explanatory variable:
corruption. The methodology consisted in using a different measurement, using the corruption score by the World Bank. The results are reported in table 7.
Although the results seem to be very similar to the ones obtained when using the main models, it should be noted that the coefficients of the corruption variable are greater in the robustness check. This can be seen as a possible support for the first hypothesis, but it fails to contribute to the validity of the model due to lack of significance.
The control variables are maintaining the same significance level and there is no variable which lost its significance when the test for robustness was conducted. This is also applicable in the reverse way, as no variable became significant in this test while it was not in the main regressions.
The similarity between the models is strengthened by the fact that no null hypothesis can be rejected by the regressions included in this test, making the end result identical to the one obtained when using the principal model.
Lastly, the goodness of fit of the models was compared and the main models are
considered to be more comprehensive based on the R-squared obtained by all of them. There is however one exception, model 2 having a better goodness of fit in the robustness test. The difference is not highly significant and the outcome of all models is the same: none can confirm either of the hypothesis.
36 Table 7: Robustness check
The following notation was used: *, **, *** for 10%, 5% and 1% significance levels
Hypothesis 1 Hypothesis 2
VARIABLES Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10
CorruptionWB -0.145 -0.170 -0.162 -0.131 -0.127 -1.000 -0.949 -0.918 -0.925 -0.890 (0.191) (0.190) (0.210) (0.209) (0.228) (0.802) (0.811) (0.850) (0.854) (0.889) CGV -0.0184 -0.0161 -0.0152 -0.0176 -0.0164 (0.0169) (0.0172) (0.0178) (0.0181) (0.0186) CWBxCGV 0.0122 0.0117 0.0114 0.0119 0.0115 (0.0091) (0.0095) (0.0098) (0.0100) (0.0103) Biddersize -0.0541 -0.0762* -0.0602 -0.0824* -0.0692 -0.0933** -0.0711 -0.0942** (0.0440) (0.0460) (0.0464) (0.0480) (0.0420) (0.0420) (0.0465) (0.0473) Leverage 0.0023 0.0020 0.0016 0.0015 0.0017 0.0014 0.0008 0.0008 (0.0033) (0.0028) (0.0037) (0.0036) (0.0034) (0.0030) (0.0039) (0.0038) ROE -0.0003 -0.0002 -0.0003 -0.0002 -0.0003 -0.0002 -0.0003 -0.0002 (0.0003) (0.0003) (0.0003) (0.0003) (0.0003) (0.0002) (0.0003) (0.0003) Quickratio -0.119* -0.125** -0.131** -0.132** -0.123* -0.128* -0.138* -0.138* (0.0642) (0.0633) (0.0661) (0.0656) (0.0684) (0.0669) (0.0715) (0.0705) Targetsize -0.0756 -0.0514 -0.0716 -0.0547 -0.0644 -0.0444 -0.0616 -0.0511 (0.0476) (0.0476) (0.0498) (0.0523) (0.0454) (0.0472) (0.0459) (0.0512) AccS 0.0013 -0.0018 0.0034 0.0009 -0.0049 -0.0082 -0.0029 -0.00549 (0.0079) (0.0082) (0.0082) (0.0081) (0.0096) (0.0097) (0.0101) (0.00971) MoP -0.865*** -0.983*** -0.984*** -1.086*** -0.722** -0.848** -0.840* -0.936** (0.316) (0.281) (0.362) (0.327) (0.362) (0.357) (0.430) (0.422) SI -0.0979 -0.0882 -0.0640 -0.0623 -0.0711 -0.0601 -0.0332 -0.0297 (0.133) (0.137) (0.140) (0.143) (0.121) (0.125) (0.125) (0.127) Constant 0.483 2.257** 2.806** 2.473** 1.869 1.794 3.880* 4.423** 4.100* 3.951 (0.335) (1.142) (1.165) (1.186) (1.338) (1.480) (2.215) (2.240) (2.279) (2.600)
Year fixed effects No No Yes No Yes No No Yes No Yes
Industry fixed effects No No No Yes Yes No No No Yes Yes
Observations 193 193 193 193 193 193 193 193 193 193
R-squared 0.009 0.080 0.106 0.156 0.177 0.049 0.120 0.145 0.189 0.210
37
5. Discussion and limitations
5.1. Discussion
The present study had the goal to investigate the relationship between BHAR and the corruption level from the target country in case of cross-border M&As. In doing so, two hypothesis were developed based on previous literature and research.
The first hypothesis assumed a negative impact of corruption on the acquirer returns measured on a three year window after the announcement date of the deal. Although the results point in the direction of the hypothesis, reporting a negative relationship between these two variables, enough evidence for supporting the hypothesis could not be found as corruption proved not significant in all the models. Using a different measurement for corruption proved the robustness of the main models, returning similar results.
Secondly, the hypothesis which aimed to test if a good corporate governance system of the bidder could overcome the negative effects of corruption and lead to positive returns was tested. This was tested by analyzing the combined effects of corruption and corporate
governance have on the returns. Again the results indicate that corporate governance can have this implication but, the evidence is not enough to draw a clear conclusion, thus leading to the hypothesis not being confirmed. The same holds when the test for robustness was performed. In testing both hypothesis, controls specific to mergers and acquisitions were accounted for by including them in the regression models. Most of the controls proved to be not significant, therefore a conclusion regarding their effect could not be reached. However, there were three that proved to be significant at times, in different models. Firstly, the acquirer size was discovered to have a negative impact on the returns, result which is in accordance with Roll’s hubris
38 also supported by previous literature (Masulis et al., 2007). Opting for a payment in cash which would remove the target shareholders signals the confidence of the acquirer in its target and its unwillingness to share future synergies. Moreover, bargaining theory suggests that in a corrupt environment the target could benefit from leverage over the bidder given the fact that it is more familiar with the environment and could ask for more of the synergies (Weitzel and Berns, 2006), case in which a cash payment is understandable.
The results lead to the following conclusions. It can be assumed that corruption is negatively affecting the acquirer returns but it cannot be confirmed until future research is performed. It appears that corporate governance of the bidder plays and important role in this equation and a developed governance system could overcome the negative implications of corruption without the firm having to resort to corruption practices or a partner. However these conclusions cannot be solidly backed due to lack of evidence.
5.2. Limitations and future research
This paper faced limitations which most likely prevented it from obtaining more
significant results which may have confirmed the two hypothesis. Perhaps the biggest limitation was related to data availability which led to a reduced sample size of only 193 observations. The fact that cross-border M&As were the subject of this research reduced the number of deals considerably. In addition, resorting to the ASSET4 database, which only includes a limited number of companies shrank the sample. Lastly, the largest number of observations was lost due to the use of DATASTREAM which mostly has data available for public firms and by the use of LaPorta’s accounting standards scores which include less than 30 countries.
Another limitation would be the fact that the bidder is originating only in the USA or UK, making possible results hardly generalizable at a global level.
