Completion or abandonment of mergers and
acquisitions in energy industries.
Master’s Thesis
International economics & Business
Supervisor
dr. P. Rao Sahib
Co-assessor
dr. T. Kohl
Johannes Jan ten Brug
1
Abstract
The energy industries experienced an increase in merger and acquisition (M&A) activity since
the mid-1990s. Based on a sample of 10.859 announced M&A attempts inward to OECD
countries during 1997–2012, this study investigates the factors that influence M&A completion
in the electricity, gas, and oil industries. The results show that divestments and privately held
target companies are key factors positively influencing the completion likelihood of an
announced M&A. Having a stake in the target company (toehold) prior the announcement, or
having experienced a number of abandoned M&A attempts in the past are the most important
factors stimulating the abandonment of an M&A deal. Related to energy industry characteristics
I find that higher levels of relatedness between the acquiring and target firms do not influence the
completion likelihood of an announced M&A, and that M&As creating national champions are
not related to a higher likelihood of completion.
Keywords: Mergers and acquisitions; completion likelihood; Energy industries;
regulation
Thanks to Padma Rao Sahib for her valuable suggestions and remarks, but most importantly for always taking time to answer questions or discuss my thesis.
2
Table of contents
1. INTRODUCTION ... 3
2. THEORETICAL FRAMEWORK ... 5
TRANSACTION LEVEL INFLUENCES ... 7
FIRM LEVEL INFLUENCES ... 11
INDUSTRY PERSPECTIVES AND REGULATORY CLIMATE ... 14
Box 1 The case of Endesa SA ... 21
Table 1: Overview of independent variables and the predicted signs ... 24
3. DATA AND METHODOLOGY... 25
DATA AND SAMPLE ... 25
DEPENDENT VARIABLE ... 26
INDEPENDENT VARIABLES ... 27
STATISTICAL METHODS ... 30
Table 2: distribution of missing values ... 32
4. RESULTS ... 33
Table 3: Overview of deals categorized by industry ... 34
Table 4: Results of logistic regression analysis ... 37
Table 5: Results of logistic regression analysis ... 38
TRANSACTION SPECIFIC VARIABLES ... 39
FIRM SPECIFIC VARIABLES ... 41
INDUSTRY AND REGULATORY RELATED VARIABLES ... 45
5. CONCLUSION ... 49
6. REFERENCES ... 52
7. APPENDIX: ... 57
TABLE 1: KAPLAN-MEIER SURVIVAL ESTIMATE TABLE ... 57
FIGURE 1: KAPLAN MEIER SURVIVAL ESTIMATE PLOT ... 58
TABLE 2: 10 LARGEST COMPLETED DEALS ... 58
TABLE 3: TOP 10 ACQUIRING COUNTRIES ... 59
TABLE 4: TOP 10TARGET COUNTRIES ... 59
TABLE 5: YEARLY DISTRIBUTION OF M&A ACTIVITY ... 60
TABLE 6.1: CORRELATION TABLE OF TRANSACTION SPECIFIC VARIABLES ... 60
TABLE 6.2: CORRELATION TABLE OF FIRM SPECIFIC VARIABLES ... 60
TABLE 6.3: CORRELATION TABLE OF INDUSTRY RELATED VARIABLES ... 61
3
1. Introduction
Fueled by technological, financial, and political changes, energy industries in most developed countries have changed drastically over the last two decades (for energy industry reviews: Verde, 2008; Jamasb and Pollitt, 2005. The structural changes made it possible for companies to obtain efficiencies through mergers and acquisitions (M&As) that were previously infeasible or prohibited under regulation (Becker-Blease, Goldberg, and Kaen, 2007). These M&As played a substantial role in the transformation of the markets, and the transformations themselves, were followed by more M&As (Andrade, Mitchell, & Stafford, 2001). Perhaps the most illustrious M&A of them all was the Royal Dutch – Shell merger in 2005, an over 80 billion dollar merger that created one of the world’s largest companies.
In the period between 1995 and 1999, which since then has been called the “fifth merger wave”, an estimated 12 thousand billion dollar was globally spent on M&As (Schenk, 2005). As the 21st century unfolded, M&As continued to play an important part in almost any industry. It has been said that at the peak of the sixth merger wave (2004 – 2008), more than 10 billion dollars a day was spent on M&As (Mathews, 2006). M&As were followed by more M&As, but also by more academic research on this topic (Haleblian, Devers, McNamara, Carpenter, and Davison, 2009; Bruner, 2004; Cartwright and Schoenberg, 2006; for overviews). Areas that attracted the most academic attention are the pre-announcement stage; studying the antecedents of M&As, and the post completion stage; studying the performance and value creation of M&As (Haleblian, 2009). One of the areas for which a considerable amount of knowledge is yet to be gained is the pre-completion stage. Despite the high costs and credibility damages related to the unilateral abandonment of a deal (Dikova, Rao Sahib, and van
Witteloostuijn 2010), not all announced M&A’s are completed, Wong and O’Sullivan (2001) estimate
that approximately a fifth of all announced M&As is eventually not completed.
4 the overall deal value was more than 370 billion dollar (Schwieters et al., 2014). In addition the energy industries are characterized by very distinctive trends. Two examples of such trends are the governmental stimulation of M&As that create large national power blocks, referred to as “national champions” (verde, 2008), and the growing amount of diversifying M&As, for example between gas and electricity firms (Jamasb and Pollitt, 2005). In this research I address the existing knowledge gap in the pre-completion stage of M&As ; with a sample of over ten thousand M&A transactions all involving at least one firm from the energy industries, I will empirically test how the completion likelihood of announced M&As is affected by industry transformations and specific factors theorized to influence the completion likelihood of these announced M&As.
5
2. Theoretical framework
The pre-completion stage of M&A activity has not been researched much. For the identification of a comprehensive set of factors on which the M&A completion likelihood is dependent, it is necessary to construct a theoretical framework. Announced M&As might be abandoned for a variety of reasons. Wong and O’Sullivan (2001) identify four broad categories in which these can be classified: voluntarily withdrawal, successful defense by the target firm’s management, rejection of the bid by the target’s shareholders, and intervention by regulatory authorities. In this study, these four groups are taken as a starting point. This allows me to identify a set of variables that might influence the completion likelihood of M&As, because the theoretical reasoning for each of these variables is based on which one of the reasons for abandonment this variable may affect.
6 Even though not much research on the pre-completion stage of M&As exists, various studies have focused on the factors that influence post M&A performance (King, Dalton, Daily, and Covin, 2004;
for a review). The factors that positively influence the post-M&A performance have a positive influence
on the economic justification of the M&A. Following the line of argument as mentioned above I expect those factors to have a negative relationship with the probability of voluntarily withdrawal and hence a positive relationship with the completion likelihood of announced M&As.
Two factors that can influence the likelihood of abandonment through successful defense strategies of the target company are the reaction of the target’s management on the takeover, and the position the target firm is in, to successfully prevent the completion of an announced M&A. Stalling or thwarting the completion process could also be important, since this could decrease the economic justification of the M&A. This could happen explicitly; for example through the costs of legal services, but also implicitly; for example through the value that could have been generated with the best alternative, if the firm had not opted for the M&A. This shows that in fact many factors might influence more than one of the reasons for abandonment, as defined by Wong and O’Sullivan (2001). The distinction between the reasons for abandonment remains important for the line of argumentation through which I identify the factors that might influence completion likelihood.
The approval of the target’s shareholders is ultimately contingent upon the received bid premium
(Loughran and Vijh, 1997). This bid premium is found to be determined by common factors in the
finance literature such as method of payment and the amount of payment related to the transaction
7 The final reason of abandonment to be considered is found in the industry specific aspects and the characteristics of the regulatory environment. These can be seen as the rules of the game for the M&A process. Differences in these “rules” in comparison with those of other industries, or rule changes for the energy industries might have a direct influence on the completion likelihood of an M&A, for example through regulatory intervention. But indirect influences on the completion likelihood should be regarded as well. An example of the latter is that the approval process of M&As in the energy industries have become longer and more tedious to ensure that the regulatory policy goals are not compromised (Coen, 2005), however this might also have increased the costs related to M&As.
Transaction level influences
M&A deals can be financed with cash, stock, debt, or any combination of those three (Haleblian, 2009). If the acquirer posts a cash financed bid, the exact value of the bid is known to the target firm’s shareholders. Stock financed M&As come with higher levels of uncertainty, since the value of the bid depends on the company’s future performance. According to the signaling theory the method of payment reflects the acquirer’s assessment of the true value of the targeted firm; bids financed with stock signal that the acquiring firm perceived the assets as overvalued, while cash financed bids signal the acquirer’s believe of undervaluation (Yook, 2003). This suggests that cash financed M&A bids are more likely to be based on economic motives, which should have a higher probability of deal completion.
8 Another factor that could affect the completion likelihood of an announced M&A is the degree of control sought by the acquirer. A higher degree of control is generally associated with a higher financial interest (O’Brien and Salop, 2000); this entails that the acquirer values the target as important. Therefore the acquirer would not want to abandon the M&A after the announcement, and the management would try its best to complete the announcement.
Contrarily if an acquirer seeks a higher degree of control, the target firm might be more averse towards the M&A, and it might have more demands during the negotiation stage since it will have less control over the firm after the M&A is completed. Additionally, M&As for which a higher degree of control is sought have to abide to more exigent requirements and regulations. These two arguments suggest that a higher degree of control sought decreases the completion likelihood of an announced M&A. Moreover, the arguments above indicate that the degree of control sought could very well have an effect on the completion likelihood of an M&A; however the direction of the effect of the degree of control on the completion likelihood is unclear for now.
9 In hostile bids the degree of the target firm’s willingness to cooperate during the M&A process is low and defensive measures that thwart the M&A are frequently used (Dawson Pence, and Stone, 1987). These defensive strategies can be initiated before and after the bid. Intuitively the completion likelihood could only be directly influenced by post bid defensive strategies. A well-known strategy is the organization of counter bids by other firms that are more in favor of the target firm’s management; the firms initiating these counter bids are called white knights (Banerjee and Owers, 1992).
Indirectly the pre-bid defensive strategies might also affect the completion likelihood of an
announced M&A, because of the relatively high price premium hostile takeovers are associated with
(Sudarsanam and Mahate, 2006). This price premium naturally affects the economic justification for the
M&A, and therefore also the completion likelihood. Often used examples of such a defensive strategies are poison pills, a strategy in which the target firm makes its stake less attractive for the acquirer. But post bid defensive strategies could also affect the completion likelihood indirectly, for example by the approval of unhealthy restructuring plans that would make the corporation unprofitable after the M&A (Sudarsanam, 1995).
To sum up, the managerial resistance of the target firms and the availability of many different defense strategies make hostile bids less likely to be completed than M&As in which the target and acquiring firm cooperate during the takeover process. This prediction is confirmed by prior research on
both completion likelihood and M&A performance (Wong and O’Sullivan, 2001; Haleblian et al., 2009;
among others), in almost all these papers, the probability of success decreases with the level of hostility of an M&A.
10 mechanisms, and large deals might be subject to closer regulatory scrutiny (Muehlfeld, 2007). Evidence on this factor is mixed; Wong and O’Sullivan (2001) find that indeed larger deals have a higher probability of being completed. Caiazza and Pozzolo (2012) find that among banks, larger transactions have a higher likelihood of being abandoned.
The share of capital already owned by a corporation in the target firm is another factor which might influence completion likelihood. A bidder’s initial holding in the target company may improve the bargaining power of the acquirer and therefore could decrease the amount of managerial resistance of the target company (Wong and O’Sullivan, 2001). However the managerial resistance could also be stimulated for M&As in which the acquirer is already an owner of the target firm. An attempt to increase one’s share is almost always a move to increase control over the target company, which consequently results in a loss of control for the target company.
An additional argument for a negative relationship is based on the “toehold effect”; a situation in which the bidder that already owns a toehold in the target company bids aggressively, as the price it offers represents not only a bid for the remaining shares but also an ask for the shares it already owns (Bulow, Huang, and Klemperer, 1999). Consistent with this argument Bulow et al., (1999) find that having an initial stake in the target company increases the chance of reaching an agreement with the target company. This seems straightforward but a side effect is that toehold bidders tend to overpay for the
companies they take over, a phenomenon referred to as the winners curse (Bajari, and Hortacsu, 2003).
11 Finally cross border M&A deals are different from domestic deals, and this might influence the completion likelihood of an announced M&A. From 1986 to 2000, 26% of the total M&A value came from cross border M&As, while in 2000, cross border M&As accounted for more than 80% of all foreign direct investment from industrialized countries (UNCTAD, 2000). In cross border deals the acquiring firms faces different cultural and institutional environments with different sets of regulatory and legal requirements, and differences in the behavior and expectations of individuals. This increases the uncertainty of the M&A process for the acquirer and there is more room for error. For example, timely and accurate information might be more difficult to obtain and interpret (Kish and Vasconcellos, 1993). Additionally foreign acquirers are often confronted with regulatory scrutiny due to protectionist policies (Kish and Vasconcellos, 1993). Therefore I expect the completion likelihood of M&As to be lower for a cross border deal.
Firm level influences
In the post-M&A performance literature target and acquirer ownership status have generally been identified as factors that influences post M&A performance (Haleblian 2009). Most research has found that M&As for which the targeted firms are privately owned performed better than M&A made on publicly owned firms (Conn, Cosh, Guest, and Hughes, 2005; Capron and Shen, 2007; Haleblian et al., 2009). This might be for a number of reasons; privately held firms are in a stronger position to defend their firm from a takeover, whenever they believe that is necessary. Moreover, since privately held firms have a stronger position during the pre-announcement stage, Muelhfeld et al. (2011) argue that “unwelcome” M&As for which the targeted firm is privately owned will have been fended off before the announcement, and therefore also have a higher probability of completion once the announced stage is reached. For publicly owned firms this is often much harder since shareholders are often dispersed and do not join forces in order to strengthen their defense.
12 possibility of information asymmetry. The textbook reaction to the risk of adverse selection is to lower the offer price (Akerlof, 1970). Lastly, M&A announcements on private companies receive far less publicity than announcements on public targets (Capron and Shen, 2007). Private target M&As might be able to “fly under the radar” until the M&A is completed, while public target M&As cannot, and might be hampered by negative publicity.
The ownership status of the acquiring firm might also influence completion likelihood. According to Shleifer (1998) private ownership is more efficient in terms of profits in industries where the needs to be innovative and to contain costs are high; characteristics that fit the energy industries very well. Kumbhakar and Hjalmarsson (1998) researched the performance of public and private firms in the electricity industry and found matching results; private firms are more efficient. This advantage in efficiency may put private firms into a better financial position to complete the M&A. Additionally, the differences in interests of owners might form a disadvantage for publicly held acquiring firms during the process of completing an M&A. Therefore I expect that announced M&As by privately held acquirers have a higher completion likelihood than if the acquirer were to be publicly held.
13 corporation on the probability of M&A completion is ambiguous; there are arguments supporting a negative relationship as well as arguments supporting a positive relationship.
A much discussed factor in the post-M&A performance literature is the M&A experience of the acquiring firm. It is easy to imagine that firms in some way learn from past experiences. Firms with M&A experience may be better prepared to select the right target firm, because they have been through the process once or more times before (Amburgey and Miner, 1992). Additionally, once the target is selected and the M&A is announced, experienced firms should be more capable to successfully move through the pre-completion stage than they were initially (Hitt, Harrison, Ireland, and Best, 1998). However empirical results have been mixed; some studies find negative effects of cumulative M&A experience on M&A performance (Hayward, 2002), others positive (Hitt, Dacin, Levitas, Arregle, and Borza 2000), or non-significant (Zollo and Singh, 2004). Few of these studies distinguish between positive or negative experience, in the pre-completion stage this distinction is more convenient to make since the performance measure is clearer (success is M&A completion, failure is M&A abandonment). Similar to Muehlfeld,
Rao Sahib, and van Witteloostuijn (2012), who find different effects for success and failure experience, I
distinguish between the contexts of experience.
14 knowledge is already high, referred to in the literature as competency traps (Levitt and March, 1988), might also cause diminishing returns for success experience.
Learning from failure is more difficult, especially when the tasks are complex (Lomi, Larsen, and Ginsberg, 1997). Therefore, given the complexities of M&A transactions, one might expect that past experiences with failed M&As do not provide the same experience benefits as successful M&A transactions of the past. Rather, I argue, similar to Muehlfeld et al. (2012) that the relationship between failure experience and completion likelihood is convex, or U shaped. Firms that have failed to complete a transaction in the past might have the wrong routines in place or might have less knowledge on how to successfully complete M&As. However firms with higher levels of failure experience might have identified these flawed routines by detecting the patterns in these past failures. Furthermore after many announced M&As that eventually had to be abandoned it seems intuitive that firms think twice before engaging in a new M&A attempt, and when they do, one would think they are fully committed.
Industry perspectives and regulatory climate
15 M&A activity is time-dependent, that is; there are so called waves in which the volume of M&A activity increases. It is likely that the M&A completion likelihood is time-dependent as well, as these merger waves are often triggered by regulatory climate changes, they may also have an effect on the completion likelihood. For the energy industries, the late 90’s have been identified as the start of an M&A wave in Europe and North America. This merger wave continued into the new millennium, in which we saw unprecedented activism of energy industry firms on financial markets (Becker-Blease et al., 2007; Jamasb and Pollit, 2005).
In 1996 the United States Federal Energy Regulatory Commission (FERC) renewed the set of M&A guidelines that limited the non-economic issues that played a role in M&A reviews. In addition to the regulatory changes, many companies had available cash flows, since technological advancements lead to lower need for capital investments in order to expand. This situation, with the help of low interest rates, lead many firms to engage in M&A activity, moreover M&A activity is seen as the means that many utility firms used in order to prepare and adapt for a competitive future (Leggio and Lien, 2000).
16 increased the completion likelihood through an improved economic rationale for M&As. The relaxed M&A regulation is also likely to have triggered a wave of M&A activity right after the passing of the act, possibly because companies were pro-actively preparing themselves for increased competition that was supposed to be a result of the Act. A point of concern for the US Government was that the increase in M&A activity could lead to market power consolidation and hence increased consumer prices. Of course the type of M&As that are suspected to cause market power consolidation will have been under tighter examination, which could have decreased the completion likelihood of these transactions.
European energy market liberalization started in the late nineties with the EU electricity directive of 1996 and the very similar EU gas directive of 1998. The directives aimed to separate the (power) generation and (retail) supply parts of the industry, from the transmission and distribution parts, which remained in their natural monopoly situation (Thomas, 2005). The reason for this separation was the ability to (gradually) open up the generation and supply markets while preventing transmission and distribution problems or instabilities (Thomas, 2005). The liberalization was twofold; first, privatization was an inevitable consequence of the directives for some countries with dominant national ownership, such as France, Italy, or Greece. Second, the cross border trading regulations and policies were revisited under the directive and transmission links between member countries were improved, mostly by EU subsidies (Jamasb and Pollitt, 2005).
17 firms, which made it unlikely that new energy generating firms would enter the market, since there would be no one to sell their power to (Thomas, 2005).
To accelerate market opening, the EU issued a second directive in 2003, this time for both the gas and electricity industries. The second directive stipulates the single energy market goal of the EU by addressing a number of specific goals, such as free entry to generation and the unbundling of transmission networks, so that companies would not use their ownership of the network to gain an advantage in their retail or generation businesses. Regardless of whether these goals were reached in time or not (the EU set July 2007 as the deadline), the directives both eased and stimulated M&A activity, since many firms reshaped their business strategies according to their expectations of the future market (Verde, 2008). An additional explanation for the rise in M&A volume is that many firms chose to opt for short term shareholder value, by acquiring rather than investing; this emphasis on short term prosperity is something that can be found in most of the highly capital intensive industries, and hence is not surprising (McKelvie and Wiklund, 2010). In addition, the rise in M&A volume was further enhanced because many European firms chose to grow inorganically, that is, through M&As, in order to withstand the competition and prevent to be taken over by larger European firms once the European market would be completely liberalized (Verde, 2008).
18
Asche, Osmundsen, and Sandsmark (2006) tested for market integration between the gas,
electricity and oil industries; they found that while gas and electricity were competing industries, the oil market was exogenously determined. However for the period after 2002 they do not find any significant results for different market behavior anymore. Serletis and Herber (1999) find that all energy industries can compete if the market structure allows it. In practice this implicates that in some place the markets are integrated while in other they are not. For example, in California, the largest US state in terms of population, the gas and electricity industries compete, while on the east coast the oil and gas markets compete.
The different types of M&As, as in inter-sectoral, horizontal, vertical, or even non-related, touches on the notion of relatedness. The relatedness between the target and acquiring firm is a factor that has been identified to influence the post M&A performance (Homburg and Bucerius, 2006). Generally, higher levels of relatedness are associated with better post M&A performance (Peltier, 2004), more related M&As are hence more likely to be based on economic motives. In the case of energy industries the economic justification might be achieved by economies of scope or scale. For horizontal transactions, M&A deals between two firms operating in the same sub-industry, the latter is the case. In addition, deal negotiations for horizontal transactions are likely to be less complicated since the structure of the target firm is more likely to be similar to that of the acquiring firm. However these type of M&As may face closer regulatory scrutiny; horizontal M&As contribute directly to market power of the firm, while the regulatory policies of both the EU and the US aimed to increase competition.
19 synergies might not be available for M&As between energy industry firms and firms from a different industry, which I refer to as unrelated inter-sectoral M&As. Therefore the completion likelihood might be different between the two types of inter-sectoral M&As. Furthermore, the completion likelihood might differ for particular types of converging M&As. Verde (2008) found that in the early 2000s many electricity companies have targeted gas companies and vice versa. By doing this, electricity companies secure supply for their plants, and gas companies secure and stabilize demand. Additionally these companies can take advantage of a broader presence in a fully liberalized market and thereby strengthen their position before the market has been fully liberalized. Altogether, the synergy opportunities appear to be more convincing for gas-electricity M&As than for converging M&As that include an oil industry firm.
An industry related point of emphasis for the completion likelihood that is similar for the US, EU, and all other OECD markets is the restructuring that comes in combination with the regulatory changes. As mentioned above, divestitures are often an important part of the market reforms in the energy industries. Governments stimulated this in order to increase competition and reduce market power of the existing firms. In the US, federally pressured divestitures of upstream business segments of state owned and vertically integrated companies plus the admission of new entrants made room for more competition at the downstream level (Kwoka and Pollitt, 2010).
20 In addition, not all divestitures are a result of restructuring. Sometimes a divestment may just be a management decision not triggered or pressured by regulatory policies. For example in June 2011, GDF Suez, a Paris, France-based provider of gas utility services, announced that it is looking to sell up to EUR 10 billion worth of assets. The investments were necessary, according to Gérard Mestrallet, CEO of GDF Suez, to “neutralize” the balance sheet and to distinguish the company from its more nuclear focused French competitors (GDF Suez SA, 2011). However this is not problematic for this research because divestments in general have a higher likelihood on being based on economic motives rather than managerial perspectives (Mulherin and Boone, 2000), which according to the logic as explained in the theoretical background section, increases the completion likelihood.
21
Box 1 The case of Endesa SA
On the fifth of September in 2005, Gas Natural SDG SA announced its acquisition of the Spanish electricity company Endesa SA; it had received the conditional acceptance of more 75 percent of Endesa’s shareholders. The acquisition was awaiting regulatory approval, while the board of Endesa was against the acquisition. Over two thirds of the total operating revenues were on the Spanish national market, therefore, the approval of the acquisition fell under Spanish jurisdiction. Endesa preferred the EU authorities to judge the acquisition because for competition reasons, the EU authorities were not likely to approve the acquisition.
In order to prevent the take-over Endesa announced on the ninth of November that it would sell some of its core power generation assets, after which more than 33 percent of the combined operating revenue is outside of Spain, this would shift the jurisdiction of the acquisition approval to the EU authorities. In the end, this attempt failed to be successful, Spanish authorities had the final say on the acquisition, most likely because the Spanish government pressured the EU authorities. The Spanish competition authority disapproved the acquisition; however the Spanish government overruled and gave its approval, probably to create a national champion that is competitive on the international market.
22
Box 1: The case of Endesa SA (Contd.)
Yet the acquisition was not completed. A German electricity company, E.ON AG, launched a rival bid that topped the bid of GAS Natural, and hence was more attractive to the shareholders of Endesa SA. Although Endesa still considered the offer of 67.1 billion to be too low, it was the higher than the bid of GAS Natural SA, and 62 percent of it would be paid in cash. The takeover, that would create the world's largest gas and electricity company with over 50 million customers in more than 30 countries around the world, was announced in February of 2006. This time the board of Endesa SA wanted the Spanish authorities to have the jurisdiction over the acquisition, the Spanish authorities were more likely to disapprove the acquisitions as indicated by a statement of Spain's Finance Minister, Mr Pedro Solbes, saying that “Spain seeks to secure a deal in its national interest in a strategic sector”. The Spanish authorities proceeded to place a set of conditions on E.ON's potential takeover of Endesa, which then lead to a formal warning of the EU that Spain’s actions were not justified. A legal battle between the Spanish authorities, who did not want to give up the set of conditions, and the EU authorities followed. At the same time Endesa proceeded to use defense tactics such as organizing other rival bids that further thwarted the acquisition. Finally, more than a year after the announcement of the acquisition, E.ON's CEO, Mr Wulf Bernotat, announced that E.ON will not carry out its takeover bid to acquire a majority stake in Endesa.
Source: Zephyr deal editorial #526197
23 standards of the EU are much more relaxed than those of the FERC in the US. Instead the EU tries to restrict companies after the M&A has taken place, for example with article 82 of the treaty of Rome, which should prevent firms to exercise market power (Notteboom, 2007). The weak power of the European authorities regarding M&As, together with the 2/3 rule, gives European national governments the ability to behave opportunistically; to push for outcomes that may seem to be in the best interest of the competitiveness of the country. However this behavior oftentimes does not serve the goal of creating a single and competitive market across the EU, a goal set by the same governments that behave in their own interest regarding the creation of national champions and the protection of national companies against the threat of being taken over by a foreign acquirer. If this theory about the creation of national champions holds empirically, domestic M&As of larger value in Europe should have a higher completion likelihood than large M&A deals outside the EU as well as smaller or cross border deals in the EU.
24
Table 1: Overview of independent variables and the predicted signs
Variables Predicted sign Argumentation
Transaction specific influences
Cash + Cash bids reflect undervaluation of the target company, which is related to the economic
justification of the M&A.
Acquired Stake +/- Higher degree of control sought could indicate that the acquirer values the target as important.
However it also represents a greater loss in control for the target company,
Recommended + The availability of defensive techniques and the managerial resistance of the target firm make
hostile bids less likely to be completed than recommended transactions
Deal value +/- Deals of larger size are organized more carefully, which suggests higher completion rates.
However, management of larger firms may be more adapt to apply defense mechanisms.
Toehold +/-
A toehold improves the bargaining power of the acquiring firm. But the move to increase control could be faced with resistance due to the loss of control for the target’s management. Also transactions for which the acquirer already had a ‘toehold’ tend to be overpaid for.
Domestic - In cross border deals the acquiring firm face different cultural and institutional environments with
different sets of legal and regulatory environments.
Firm specific influences
Public target - Privately owned firms tend to outperform public firms, and private firms tend to be traded at a
price discount due to difficulty in information gathering.
Public acquirer - The higher efficiency of private firms may put them in a better position to complete M&As.
Dispersed interest in owners of public firms may also be a reason for lower completion likelihood
Target is subsidiary - The existence of the ultimate owner forms an extra layer, with its own strategic agenda in the
negotiation process
Success experience +
Successful M&A experience could indicate that those firms do something right during the process of acquiring. Furthermore firms can use these past experiences by encoding and fine-tuning routines and knowledge for future M&A attempts.
Success experience2 - The knowledge that can be gained diminishes for every subsequent M&A transactions since this
knowledge is relatively similar.
Failure experience - Firms that have failed or abandoned an M&A transaction in the past may have the wrong routines
in place or might have less knowledge on how to successfully complete M&A deals.
Failure experience2 + Firms with high levels of failure experience might have identified flawed routines.
Industry related influences
Divestment +
Divestments in general have higher likelihood on being based on economic motives. Furthermore, divestments are stimulated or incentivized by regulatory bodies to restructure the markets, and hence face less constraint of these regulatory bodies.
Horizontal +/-
Higher levels of relatedness are associated with better post M&A performance, and hence higher completion likelihood. In Horizontal M&As relatedness is high. However they might also face high regulatory scrutiny because of market power consolidation arguments.
Vertical +/-
Synergies might make these M&As more favorable. However the regulatory scrutiny also is an issue for the type of M&As, the unbundling of energy industry tasks has been an objective of regulatory bodies during times of restructuring.
Convergence +/- Similar issues as the above two variables. However the influence of on the one hand synergies and
25
3. Data and Methodology
Data and sample
For this study I have collected all M&A transactions inward to OECD countries, irrespective of their completion or country of the acquirer, for which at least one of the firms involved was in the electricity industry that occurred between January 1, 1997 and January 1, 2012. Currently (June, 2014) the OECD consists of 34 member countries, a list of which can be found at the OECD website (oecd.org), countries that joined the OECD after 1997 are nonetheless included in the sample. The M&A data are derived from the Zephyr Database, which includes worldwide M&A data dating back to 1997. In order to prevent a sample selection bias, deals that were announced in 2012 and completed in 2013 are included, even though 2013 is technically beyond the scope of my time sample. Transactions that are announced later than December 31, 2012 are not included in the sample. Limiting the research to only OECD countries gives me the opportunity to emphasize on the European Union and North America energy markets, two regions with high volumes of M&A activity that underwent industry changes over the course of the time sample of this research. Furthermore the OECD countries account for the majority of M&A activity. Of all completed deals reported in the Zephyr database over the course of my time sample, 82 percent targets a firm within an OECD country. Furthermore, companies from OECD countries accounted for 87% of inward cross-border M&As in 1991-98 (Kang and Johansson, 2000).
Only true merger or acquisition transactions are included; initial public offerings, repurchases of own shares, or self-tender offers are excluded. This yielded a total of 10.859 transactions of which 9.136 (84,16%) were eventually completed. A similar completion percentage was found in other M&A completion studies (Muelfeld et al., 2010; Dikova et al., 2010).
26 combination of firms, for these observations the accompanying firm level data is collected for the largest of the firms. The total number of observations with this characteristic was not large enough to test whether this has an influence on the completion likelihood; instead I treat them as any other observation. In addition, for the majority of these cases the group of firms had the same ultimate owner. The accompanying firm data is collected from the Orbis database. Both databases are provided by Bureau van Dijk, which helps the possibility and convenience of merging the two data sets.
The industry classification used in this research is the primary NACE code 2nd revision, which indicates in what industry the primary business of the firm is based. Transactions for which either the target company or acquiring company has a three digit primary NACE code in one of the following energy industries are selected: 061 (Extraction of crude petroleum), 062 (Extraction of natural gas), 091 (Support activities for petroleum and natural gas extraction), 351 (Electric power generation, transmission and distribution), 352 (Manufacture of gas; distribution of gaseous fuels through mains). I further distinguish sub-industries on the fourth digit NACE code, in order to test the influence of relatedness on the completion likelihood.
Dependent variable
The dependent variable, M&A completion, is measured through a dummy variable that takes the value one if an announced deal is completed and the value zero if an announced deal is later abandoned.
The date of announcement is used as the start date of the transaction and the date of completion or abandonment as the solution date. In case of a merger, the largest partner is seen as the acquiring firm. There are some announced transactions for which the database does not provide information on whether the transaction is completed or abandoned1. For these transactions a cutoff point has been determined through a nonparametric event history analysis, which yields a Kaplan-Meier survival estimate below five percent for 225 days or more. That is, for transactions that have not been either completed or abandoned
1
27 for 225 days or more, there is less than a five percent chance that they still will be completed. Therefore, transactions for which the solution (completion or abandonment) is unknown are considered as abandoned if the last update was more than 225 days ago (as of march 31, 2014). The unknown outcome observations that were announced within 225 days of March 31, 2014 have still a reasonable chance of being completed and are therefore dropped from the sample.2 This was the case for a total of five transactions. The results for the Kaplan-Meier estimation are shown in table 1 and figure 1 of the appendix.
Independent variables
The independent variables in this study can be organized into three groups; transaction specific characteristics, firm specific characteristics, and industry specific characteristics. In this subsection the independent variable will be outlined following these three groups. Names of the variables are presented in italics.
The first transaction specific independent variable related to the method of payment. As described in the theoretical framework I hypothesize that transactions paid in cash have a higher likelihood of completion than transactions paid in debt or stock. In order to test this effect a dummy variable is created. This variable takes the value of 1 if at least 51 percent of the transaction is paid in cash and 0 if otherwise. Transactions that are financed by a combination of stock or debt and cash for which the share of cash does not exceed 51 percent make up the reference category.
Degree of control sought is a variable that captures the percentage of (common) shares in the
target firm sought by the acquiring firm. For the variable capturing the acquisition attitude a dummy is created that takes the value 1 if the transaction is a so called recommended bid and 0 otherwise. A transaction is considered recommended if the management board of the target company has agreed that the terms of the takeover bid are fair and have recommended that their shareholders accept the offer. The
2
28 opposite extremes of recommended bids are hostile takeovers (Zephyr, 2014), but because not all transactions are either hostile or recommended, non-recommended bids make up the reference category.
The next transaction specific variable, deal value, is measured in tens of millions of Euros. The largest transaction by deal value in the sample is the failed acquisition attempt of ELECTRICITÉ DE FRANCE SA (FR) on E.ON AG (DE) in 2008. The reported value of the bid was 83,875 billion euros, had the deal gone through, this would have been the third largest M&A deal of all time (Bloomberg, 2013). Table 2 in the Appendix provides an overview of the 10 largest completed deals of my sample.
The toehold dummy tests whether the share of capital already owned by a corporation in the target firm influences the completion likelihood. It takes on the value of 1 if the acquiring corporation possessed a given percentage of shares (common or common equivalent) of the target company, before the announcement of the current transaction. If the acquirer was not an already an owner of the target firm, the toehold dummy takes the value of 0. The final transaction specific variable is domestic transaction. This dummy variable takes the value 1 for a transaction that is domestic; when the target and acquiring firm reside in the same country, and a value of 0 for transactions between firms from different countries.
The research includes four independent variables in the firm characteristics category. The first,
public acquirer is a dummy variable that takes the value 1 when the acquiring firm is publicly owned and
the value 0 for non-publicly held firms. Public target is included for the ownership status of the target firm, measured in the same way as the public acquirer variable. The subsidiary status of the target firm is the third firm specific dummy variable, that measures whether the targeted firm is a subsidiary of a larger corporation (= 1) or not (= 0).
29 transactions are counted, and failure experience, for which prior abandoned transaction are counted. The predicted relationship for both types of experience is non-linear; therefore the quadratic terms are added as separate variables.
Out of the 4380 acquiring firms in the sample, there are 330 firms3 that attempted at least 5 transactions, and 2884 firms that only attempted to acquire once. Even though there are more M&A transactions initiated by North American firms in the sample, EU firms seem to have more M&A experience. The average success experience for EU firms is 3.65 and their average failure experience is .7. The average success experience for North American firms is 2, and their average failure experience is only 0.13. In the EU, the five percent firms with the highest success experience have all successfully merged or acquired at least 19 times, for failure experience, the 95th percentile score is 4. In North America, the 95th percentile scores are 9 for success experience and only 1 for failure experience. The percentage of M&As initiated by a firm without any success experience is similar between firms from the EU (43.6%) and North America (45,3%). For failure experience these percentages differ more between the two regions, 25 percent of the acquirers from the EU already experienced a failed M&A attempt, while only 9 percent of the North American acquirers experienced a failed M&A attempt.
In order to research the impact of industry features this study contains a number of industry related variables. Divestment is a dummy variable that captures whether the target firm divests part of its assets or a subsidiary (=1) or not. No distinction is made between the reasoning behind the divestments.
The variables horizontal, vertical, and convergence test the notion of relatedness. The relatedness of an M&A refers to the level of diversification between the target and acquiring firm and thus the likelihood of possible synergies between the firms. A transaction is horizontal (=1) if the target and acquiring firm are operating in the same sub-industry, and non-horizontal (=0) if not. A transaction is
vertical (=1) if the target and acquiring firm are operating within the same industry (NACE code, 3rd
3
30 digit) but in a different sub-industry (NACE code, 4th digit), and non-vertical (=0) if not. Finally,
convergence is made up out of transactions between the three industries (NACE code, 3rd digit) (=1). The observations for which this is not the case get the value of 0. Transactions between firms from one of the energy industries and a firm from an unrelated industry are not categorized as converging transactions. This last category makes up the reference group, and allows me to enter all three variables into the model without falling into the dummy variable trap.
Lastly the time dependency of M&A completion likelihood is tested by three dummy variables, which are specific to a particular industry. The first is for the Gas and electricity industry and captures the period between 2001 and 2003, when the market was characterized by price instabilities and recurring large scale blackouts. Also for the gas and electricity industries, I define a post-2003 indicator, because this period represents a structural shift in the regulatory climate of both the US and the EU. Finally, in the oil industry model, a dummy variable for the period between 2004 and 2008 is added, since this represents a structural increase in global oil prices.
Statistical methods
An extensive analysis of the descriptive statistics of the data helped finding patterns in the energy industries for the different geographical areas in my research. In addition I run a series of binary logistic regression models with M&A completion as the dependent variable to test for the effects of the explanatory variables on the completion likelihood of an announced M&A. The model is written as follows:
31 The dataset contains M&A transaction data over time, however I do not follow the behavior of specific firms over time; hence it is not a panel dataset. Nonetheless serial correlation might be a potential problem since some firms in my sample have engaged in more than one transaction. To control for this within firm correlation standard errors are clustered by acquirer, using the “Huber sandwich estimator” (Freedman, 2006).
For some of the independent variables there is little data available, which resulted in many missing observations in these variables. For these variables a missing-indicator method is used (Cohen and Cohen, 1983); missing data dummy variables, that take the value 1 when the observation from the original variable is missing, and 0 when this is not the case, are added to the model. Subsequently, the missing observations in the original variables are replaced by zeroes. By so doing I still test for the effects of these variables on the completion likelihood, while not losing the benefits of the large sample size for the remaining variables. There are no indications that the missing values are selective4. That is, the distribution of missing cases appears to be non-related to the dependent variable. The inclusion of significant but in essence meaningless missing-indicator variables could lead to an overestimation of the ROC area (Van der Heijden, Donders, Stijnen, and Moons, 2006), which implies that the model overestimates the accuracy of the results for that particular variable (Copas and Corbett, 2002). Van der Heijden et al. (2006) empirically test this method in logistic regressions and in spite of the overestimation issue, the missing-indicator method nonetheless produced better results than an analysis of the non-missing values only. The non-missing-indicator method is used for the variables relating to the sub-deal type (acquisition attitude and divestment), the method of payment variable, and the deal value variable. Table 2 shows the number and percentage of missing observations for each of these variables.
4
32 I will estimate 7 models; shown in table 4 and 5, further on in this paper. The first model tests for all independent variables and includes all M&A announcements, this is the benchmark model. Models 2, 3, and 4 test for the completion likelihood on the subsamples for which the target firm is from the EU, North America, or the “rest of the world” (RoW). The distinction between the three regional areas is made because of the large differences in the regulatory environments as indicated in the industry analysis. Table 5 shows the last three models; model 5 tests for the completion likelihood in the electricity sub-sample, and includes all transactions for which the target firm operates in the electricity industry. Models 6 and 7 show similar models as model 5 but then for the gas and oil industry subsamples respectively. A pooled regression is run for the three different industries in the benchmark model. Separate regressions are shown in models 5 through 7 because of the inconclusive evidence to what extent these markets are integrated, and compete with each other (Asche et al., 2006; Serletis and Herber, 1999).
In addition an ANOVA test shows that the completion likelihood is not the same for all regions or industries, in fact a Bonferroni pairwise test indicates that the completion likelihood is statistically different between all groups. Furthermore I test whether the coefficients of the different variables are the same across groups. Allison (1999) pointed out that conventional tests of the equality of coefficients across groups, such as the chow test, could be problematic in binary regression models, because such tests mix up the magnitude of the regression coefficients with the residual variation in logistic or probistic models (Long, 2009). The ordinal generalized linear model as proposed by Williams (2010) solves this issue, and yields that there are considerable differences between coefficients of different regions or
Table 2: distribution of missing values
Variable name known (out of 10.859) missing (out of 10.859) % missing
Recommended 2.406 8.453 77,84
Divestment 2.406 8.453 77,84
Cash 4.527 6.332 58,31
33 0 200 400 600 800 1000 1200
N
umber
of T
ransac
ti
ons
Year
Figure 1: Volume of M&A deals by year
Total Attempted M&As Total completed M&As
industries. By and large, these tests strengthen my beliefs that it is wise to test for each model separately, in addition to the model that includes all observations.
4. Results
34 Figure 1 represents the annual number of M&A attempts and the number of completed M&As per year. An examination of this graph shows that the total number of M&As rose consistently and rapidly between 1997 and 2005. This is consistent with findings of previous literature which showed M&A volume growth rates as high as 80 percent in the second half of the nineties (Bruner, 2011), and that the
growth rates in the energy industries spurred around 1997 (Verde, 2008). Accompanying the rise of total
M&A attempts is the decrease in completion rates, which were substantially higher from the start of the sample until 2001. Table 5 of the appendix displays these completion ratios for each year of the sample.
Table 3 shows the number of transactions initiated per 3 digit NACE code; for all industries except the support activities for oil and gas the highest number of transactions was attempted within the same industry. Some surprising observations can be noticed; the low number of M&A attempts of either gas or oil industry firms in the oil and gas support activities industry. The electricity industry is the only industry in the sample that has more incoming than outgoing transactions, and the most transactions were attempted in the electricity industry. Transactions for which the target firm is in the same industry as the acquiring firm can either be horizontal or vertical; this sample includes 2,452 horizontal transactions (25.5% of total), and 1,609 vertical transactions (16.8% of total). Transactions between firms categorized in the support activities for oil or gas and firms from either the oil or gas industry are considered vertical.
Table 3: Overview of deals categorized by industry
Acquirer industry
Unrelated Electricity Gas Oil Support act.
for oil & Gas Total
T arg et ind us try Unrelated 0 1,026 244 337 537 2,144 (22.3 %) Electricity 1,068 2,274 131 50 10 3,533 (36.9 %) Gas 298 232 342 65 13 950 (9.9 %) Oil 847 47 60 619 146 1,719 (17.9 %) Support activities
for oil & Gas 652 22 28 135 413 1,250 (13 %)
Total 2,865 (29.9%) 3,601 (37.5%) 805 (8.3%) 1,206 (12.6%) 1,119 (11.7%) 9,596 (100 %)
35 coefficients of all right-hand side variables are documented in the appendix (table 6). There are high correlations between the explanatory variables for which missing indicator method is used (Cash, Recommended, Deal value, and Divestment) and their corresponding missing indicator variable. This is by construction, because the values for the missing indicator variable are determined by whether these observations were missing in the original variable or not. Therefore a larger number of missing observations in the variable for which the MI method is used, results in a higher correlation between that variable and its corresponding missing indicator variable. The largest correlation coefficient among the variable for which the missing indicator method is used is 0.65.
The correlation coefficient between the independent acquired stake variable and the initial stake dummy variable is 0.858. The variance inflation factors (VIFs) are 4.03 and 4.2 (for the initial stake dummy-variable and the acquired stake variable respectively), well below the often used cutoff level of 10 and also below the more conservative cutoff point of 5 (Hair, 2009). Hence I do not apply any correcting measures. Lastly the experience variables are highly correlated with their squared term variables, the accompanying VIF levels for success experience and its squared term are 7.37 and 6.45 respectively. For the failure experience variable and its squared term these are 5.67 and 4.32 respectively. However since the inclusion of the squared term variables is necessary to test for diminishing returns, I refrain from applying correcting measures.
The studentized residuals, deviance residuals, and their leverage have been analyzed for outliers. Using a cutoff point for the studentized and deviance residual of 3 and a leverage cutoff point of 2p/n5, yields a total of 26 cases. These outliers are listed in the appendix and have been examined separately. Among these there are five observations of acquisitions initiated by Dutch firms that report some unexpected values: the target and acquiring firm are the same, and the acquired stake is zero. After examining these companies it appears that these transactions had only legal or accounting purposes, and therefore these are excluded from my empirical analysis. Furthermore, among the 26 outliers there are
5
36 two transactions for which I find duplicate observations in the sample. In both cases these observations are excluded from the empirical analysis. For the remaining cases there are no indications why these should be excluded from the sample. It is noteworthy that 6 of the cases are among the largest deals in terms of deals value (greater than 2 billion) and that all but one are abandoned deals. The exception being the Shell - Royal Dutch merger, which is also the largest observation of the dataset, in terms of deal value.
Table 4 and 5 contain the regression models together with the value of the pseudo likelihood function6. The null hypothesis of the Wald chi square test, that all coefficients associated with the independent variables are equal to zero, could be rejected for all models. The legend of each regression table can be used while interpreting the significance of the coefficients. The coefficients of the variables are displayed in log-odds. These raw logistic coefficients should be interpreted with caution since they refer to the increase in logarithmic odds resulting from a one unit increase in a variable, while each variable has a different underlying measurement scale. For the reader’s convenience, I will refer to the marginal effects whenever the magnitude of the influence of an independent variable is discussed. The marginal effects show how the dependent variable, the completion likelihood of an announced M&A, is changed by an additional unit, holding all other variables in the equation constant (Hair, 2009).
6
37
Table 4: Results of logistic regression analysisᵀ
(1) (2) (3) (4)
Variable name overall sample Target EU Target N-America Target RoW
Transaction specific variables
Cash 0.321** 0.178 0.244 0.291
(0.131) (0.234) (0.189) (0.295)
Degree of control sought -0.0109*** -0.00841** -0.00873 -0.0138**
(0.00289) (0.00378) (0.00816) (0.00556) Recommended 0.636*** 0.716 0.563** 1.506*** (0.185) (0.504) (0.246) (0.437) Deal value -0.00765*** -0.00816** -0.00784*** -0.0103* (0.00213) (0.00388) (0.00285) (0.00593) Toehold -0.989*** -0.729*** -1.193** -1.433*** (0.187) (0.257) (0.491) (0.367) Domestic 0.366*** 0.217* 0.545*** 0.672*** (0.088) (0.123) (0.183) (0.191)
Firm specific variables
Public target -0.932*** -0.819*** -1.890*** -0.913*** (0.110) (0.131) (0.324) (0.241) Public acquirer -0.0343 0.0247 -0.096 0.128 (0.0934) (0.113) (0.388) (0.254) Target is subsidiary -0.293*** -0.420*** 0.0311 -0.387** (0.0853) (0.121) (0.178) (0.196) Success experience 0.115*** 0.1015*** 0.134*** 0.150** (0.0215) (0.0218) (0.0489) (0.0647) Success experience2 -0.0031*** -0.0032*** -0.00127 -0.00835** (0.000784) (0.00094) (0.00218) (0.00351) Failure experience -0.5313*** -0.433*** -1.056*** -0.483** (0.0799) (0.0902) (0.337) (0.227) Failure experience2 0.0423*** 0.0339*** 0.148 0.0695 (0.00949) (0.009) (0.107) (0.0447)
Industry related variables
Divestment 2.210*** 1.603*** 2.724*** 3.337*** (0.290) (0.379) (0.743) (0.667) Horizontal 0.0024 -0.0975 0.0448 0.134 (0.0884) (0.130) (0.170) (0.211) Vertical -0.3069*** -0.328** -0.183 -0.565** (0.102) (0.149) (0.218) (0.218) Convergence -0.1842 -0.387** -0.0457 0.251 (0.140) (0.175) (0.308) (0.408) Target in EU 0.334*** (0.105)
Target in North America 0.445***
(0.113) 2001 - 2003 -0.840*** -0.572** -1.088*** -1.042** (0.179) (0.249) (0.346) (0.445) After 2003 -0.706*** -0.486** -0.990*** -0.899** (0.171) (0.232) (0.322) (0.427) Constant 2.829*** 3.253*** 3.044*** 2.979*** (0.356) (0.481) (0.909) (0.778) Observations 7,751 3,167 3,396 1,188 Pseudo R-squared 0.134 0.118 0.119 0.189 Log pseudolikelihood -2419 -1157 -802.3 -419.5
Robust standard errors in parentheses
38
Table 5: Results of logistic regression analysisᵀ
Variable name
(1)
overall sample electricity Ind.(5) Gas industry(6) oil industry(7) Transaction specific variables
Cash 0.321** 0.444** 0.217 0.0093
(0.131) (0.237) (0.291) (0.195)
Degree of control sought -0.0109*** -0.00630 -0.0103 -0.0232***
(0.00289) (0.00446) (0.00711) (0.0061) Recommended 0.636*** 1.294*** 0.542 0.968*** (0.185) (0.452) (0.385) (0.248) Deal value -0.00765*** -0.0140*** -0.0122*** -0.00286* (0.00213) (0.00376) (0.00385) (0.00157) Toehold -0.989*** -0.632** -0.904** -1.875*** (0.187) (0.304) (0.461) (0.353) Domestic 0.366*** 0.165 0.611*** 0.522*** (0.088) (0.144) (0.189) (0.170)
Firm specific variables
Public target -0.932*** -1.113*** -0.505** -0.674** (0.110) (0.157) (0.256) (0.285) Public acquirer -0.0343 0.0861 -0.144 0.257 (0.0934) (0.144) (0.234) (0.262) Target is subsidiary -0.293*** -0.491*** -0.399** 0.097 (0.0853) (0.136) (0.184) (0.262) Success experience 0.115*** 0.105*** 0.0568 0.0787 (0.0215) (0.0331) (0.0509) (0.0635) Success experience2 -0.0031*** -0.00328*** -0.00257 -0.00539 (0.000784) (0.00113) (0.00169) (0.0036) Failure experience -0.5313*** -0.424*** -0.243 -0.698*** (0.0799) (0.121) (0.181) (0.159) Failure experience2 0.0423*** 0.0373*** 0.0151 0.0805*** (0.00949) (0.0129) (0.0276) (0.022)
Industry related variables
Divestment 2.210*** 2.414*** 2.272*** 2.838*** (0.290) (0.447) (0.629) (0.759) Horizontal 0.0024 -0.123 -0.0265 0.110 (0.0884) (0.156) (0.245) (0.186) Vertical -0.3069*** -0.469*** -0.151 0.151 (0.102) (0.169) (0.238) (0.247) Convergence -0.1842 -0.267 -0.261 0.437 (0.140) (0.266) (0.238) (0.451) EU 0.334*** 0.244 0.554** 0.378* (0.105) (0.167) (0.233) (0.225) North America 0.445*** 0.203 0.892*** 0.296 (0.113) (0.198) (0.236) (0.203) 2001 - 2003 -0.840*** -0.774*** -1.430*** (0.179) (0.286) (0.464) After 2003 -0.706*** -0.556** -1.019** (0.171) (0.276) (0.439) 2004 - 2008 0.674*** (0.158) Constant 2.829*** 2.629*** 3.062*** 2.807*** (0.356) (1.126) (0.895) (0.695) Observations 7,751 2,773 1,824 2,465 Pseudo R-squared 0.134 0.153 0.151 0.144 Log pseudolikelihood -2419 -958.8 -511.7 -677.1
Robust standard errors in parentheses
39
Transaction specific variables
The right-hand side transaction specific explanatory variables in the models are method of payment, degree of control sought, recommended transaction, deal value, the toehold dummy and
domestic. The cash payment variable had a positive effect in the overall sample; however in none of the
geographic sub-regions the variable is significant. The only significant coefficient is found in the electricity industry; M&As for which a firm from the electricity industry is targeted, have a significant higher completion likelihood if at least 51 percent of the deal value is paid in cash.
These findings are in contrast with those of Muehlfeld et al. (2007), who find very robust indications that cash transactions are related to higher completion rates. It could be that the value assessment for firms from energy industries is relatively straightforward, and easier the value assessment for firms from other industries. Therefore the theory that a cash payment signals undervaluation and a debt or stock payment signals overvaluation is of relevance for M&A attempts made on firms from the energy industries. The average deal value of M&As in the energy industries could also be of influence. With an average deal value in this study of 572 million, I find that the deal value of M&As from energy industries are generally larger than transactions in most other industries (Schwieters et al., 2014). A larger deal value might alter the decision to pay at least 51 percent of the deal value in cash, which might be an explanation for the difference in findings between this study and some of the prior completion likelihood studies.
