This paper is first to empirically examine the influence of the degree of leverage on cross-border acquisitions

Investigating the influence of leverage on acquisitions.

Evidence from the UK

Rudolf Bakker^*

Supervisor: dr. J.H. (Henk) von Eije Thesis: MSc Finance and MSc IFM

June 2016

Abstract

This study investigates the influence of leverage on acquisitions. This paper is first to empirically examine the influence of the degree of leverage on cross-border acquisitions.

I use panel data on 3900 United Kingdom listed firms over the 2000-2015 period.

Firstly, the results show evidence for a negative relation between leverage and acquisitions. Secondly, I find slight evidence implying that firms engaging in cross- border acquisitions have lower leverage levels than firms that undertake domestic acquisitions. The results have implications for policy makers and practitioners regarding the timing and degree of leverage prior to acquisitions.

JEL classification: G30, G34

Keywords: Acquisition, Leverage, Capital Structure

* Faculty of Economics and Business, University of Groningen, PO BOX 800, 9700 AV Groningen, the Netherlands. Student number: s1918508; email: rudolfbakker@hotmail.com

2

1. Introduction

In the last decade historically high stock prices and a financial crisis with an enormous impact on the world economy were observed. However, preceding the financial crisis leverage levels were relatively high and rising (Reinhart and Rogoff, 2008). At the same time the sixth merger wave was observed (Martynova and Renneboog, 2008). These observations give rise to several questions. One of them might lie in the relation between rising debt levels and the documented merger wave. Although existing literature on mergers and acquisitions are extensive yet, the influence of capital structure on takeover decisions is not deeply researched recently.

In previous studies the probability and the likelihood of engaging in acquisitions is investigated however, results remain ambiguous. Uysal (2011) investigates whether managers deviate from their target capital structure in order to be able to undertake acquisitions. He finds that overleveraged firms relative to their target debt ratios are less likely to make acquisitions and are less likely to fund their acquisitions with debt.

Furthermore, Vermaelen (2014) addresses the importance of access to debt markets on investment decisions. He stresses that rated firms are more likely to undertake acquisitions than nonrated firms. However it remains hard to find conclusive results whether the degree of leverage is indicative of the firms’ ability to undertake acquisitions.

In this paper I study the effect of leverage on takeover decisions. In particular, I investigate using a sample of UK listed acquirers and non-acquirers, whether the degree of leverage in a particular firm is indicative for its takeovers. I employ a probit model using various leverage measures in order to examine the impact of leverage on the likelihood of takeover decisions. The model presents two testable hypotheses. The first hypothesis is that leverage is negatively associated with the probability of takeover. The second hypothesis is that firms that engage in cross-border acquisitions have lower debt levels than firms that engage in domestic acquisitions. In order to test the hypotheses I use an UK sample consisting of 3900 listed firms. The sample contains firms that acquire and firms that do not acquire.

This paper adds to the existing literature in several ways. Firstly, this paper re- examines the work of Uysal (2011) and Harford and Uysal (2014) by re-examining the effect of leverage on acquisitions. Uysal (2011) investigates the impact of firms

2 deviating from their target debt level in order to undertake acquisitions, whereas Harford and Uysal (2014) investigate whether having a debt rating increases the likelihood of acquisitions. Both use probit regression models and take financial ratios into account in order to determine the likelihood of acquisitions. Secondly, I delve deeper into the leverage discussion and its effects on acquisition decisions. I do this because leverage and acquisitions are among the most important topics in finance and yet there is still no conclusive research on their relation. Another study from the perspective of acquires is the study of Bruner (1988). His study finds a large increase of leverage among bidders in the year of merger and therefore he hypothesizes that levering is associated with the purchase of targets. Furthermore, Shimizu et al. (2004) addresses that there exists a lack of proper research on many areas regarding globalization and cross-border acquisitions. Although many studies investigate determinants of cross-border acquisitions and their returns, little research has addressed the influence of leverage on cross-border acquisitions. With this study I try to fill this gap.Thirdly, I add to the existing literature by comparing the relation of leverage on acquisitions in a domestic setting as well as a cross-border setting. Lastly, to my best knowledge this is the first study to empirically examine the degree of leverage in firms and whether this degree of leverage is indicative for undertaking (cross-border) acquisitions using a sample consisting of acquirers and non-acquirers.

The empirical results of this study provide support for the first hypothesis. I find a statistically significant negative relation between a firms’ leverage and its engagement in acquisitions. This result indicates that firms that are relatively under levered are more likely to undertake acquisitions. Evidence for the second hypothesis is also documented, however one has to be cautious when interpreting this result. I find a negative relation between leverage and acquisition where the coefficient for the foreign sample is more negative compared to the domestic sample. This result implies that firms that acquire cross-border do have lower leverage compared to the firms that engage in domestic acquisitions. The results provide relevant insights for both academics and practitioners in a sense that leverage is always an important issue to managers. Managers constantly need to reconsider whether their firm obtains the optimal capital structure. Especially when it comes to acquisitions this study provides relevant information.

The remainder of this paper is structured as follows. In the next section I review the existing literature and present the hypotheses analyzed in this study. The third

3 section consists of the data and variables used. Section four provides the methodology and section five concludes.

2. Literature review & hypotheses

In this section the literature regarding the existing theories on leverage and acquisitions are discussed. Hereafter, I present an overview of the empirical literature regarding the effect of the degree of leverage within acquirers on acquisitions. Followed by stating the hypotheses as related to the theory and the empirical literature.

2.1 Theory

Here, I elaborate on the most important existing theories regarding leverage and acquisitions. Modigliani and Miller (1958) were on the basis of the capital structure discussion. Their study stresses that in a world free of taxes, bankruptcy and market imperfections a firm is independent to the level of equity and debt it issues. Hereafter, I describe the theories that build further upon the Modigliani and Miller (1958) study, since in practice there exist taxes and markets seem not to be that perfect. In the following paragraphs I discuss theories that relate to these issues as for instance, tax shields, the agency theory, bankruptcy costs and debt overhang.

In a world where taxes exists, issuing debt in exchange for stock creates a tax advantage since the interest payments on debt are tax deductible. Therefore a tax shield occurs when a firm takes on debt (Modigliani and Miller, 1958). According to their theory the amount of debt or equity in a firm is independent to firm value in a world without taxes. However, when considering a world where taxes exist there is a tax advantage to be reaped for firms a so-called tax shield. If a firm takes on debt the interest payments the firm makes to the lender are tax-deductible. So in this case the firm can increase its firm value by taking on debt. Although this situation would imply that the capital structure of firms should consist entirely of debt, this situation is in reality hardly unlikely.

This is known as the bankruptcy (costs) theory. Bankruptcy costs occur when firms exceed their optimal capital structure. Firms have optimal capital structures where the benefits of the tax advantage and the costs of debt are offset. As firms increase their debt beyond the optimal capital structure, it becomes riskier to lend for these firms

4 and therefore interest payments made to the lender will increase. The increase in the cost of debt will (eventually) affect the overall firm value. Scott (1976) describes this relation and states that under the assumption that bankruptcy is zero, the model is almost identical compared to the Modigliani and Miller (1958) study. However if there is a possibility that a firm can go bankrupt there exists an unique optimal capital structure which is not only dependent upon expected future earnings but also on the liquidating value of its assets. The framework is primarily based on the situation that interest payments are tax deductible. It is detrimental to the extent that increases in the level of debt increase the probability that the firm will incur bankruptcy costs.

Another theory relating to the determination of debt and equity levels is the agency theory. On the basis of this theory is Jensen (1986), who thoroughly described the influence agency costs have on determining debt and equity levels. The agency theory describes the conflicts of interest that arise between the owners and the agents of the firm when incentives and goals are not aligned. Corporate managers are the agents of the firm and shareholders are the owners of the firm. Shareholders strive for shareholder value maximization obtained through the highest share price possible, whereas agents might have incentives that will decide them to take actions that are not optimal for the shareholders with regards to shareholder value maximization. In particular, managers might behave in their self-interest in order to obtain financial rewards or to obtain more resources under control (Jensen, 1986). According to Jensen (1986) especially firms with high free cash flows face agency problems. Managers may rather use cash to invest in projects than paying out to shareholder or buyback shares.

Instead they might want to exert their influence and invest the free cash flow in low or even negative NPV projects and other organizational inefficiencies. Therefore firms want to discipline managers to prevent them from undertaking value destroying NPV projects. Jensen (1986) explains how using debt can reduce agency costs. In other words firms use debt to discipline managers and control their actions. The author refers to this as the “control hypothesis”. In this way debt is used to motivate managers and their organizations to become more efficient.

However increasing leverage also has costs and caution must be taken into account that debt issuing not always works. If a firm faces too excessive debt levels it faces debt overhang problems. This theory is described by Myers (1977) and is the fourth theory described in this study on the relation between leverage and investments.

5 Thus, the debt overhang describes that there exist limitations to corporate borrowing.

Firms with risky debt outstanding and acting in the interest of the shareholders will follow different decision rules than firms that can issue risk free debt or issue no debt at all. This clearly indicates that firms facing debt overhang are worse off. Therefore firms should not issue debt above their optimal capital structure level.

Myers and Majluf (1984) relate the agency theory of Jensen (1986) to the asymmetry theory and state that managers have monopolistic access to firm-specific information creating information asymmetry between in and outsiders. Furthermore, they indicate that managers can sufficiently gain from acquisitions when the managers believe that their securities are (incorrectly) overvalued by investors. Therefore leverage of firms might be a restriction or an opportunity for firms when facing acquisitions decisions.

Additionally, Jensen and Meckling (1976) document moral hazard an explanation of agency cost of debt. High levels of debt might induce firms to opt for excessive risk taking and thus taking on risky projects. The authors illustrate that there is almost certain no creditor willing to invest 100 million dollar in a firm if the entrepreneur has invested only 10,000 dollar. The probability that the entrepreneur engages in more risky behavior (e.g. undertaking more risky projects) ex ante is very likely, since in this case the entrepreneur faces upside potential and limited downside risk. In case of default the creditor loses money leaving the entrepreneur with only a loss of 10,000 dollar. However, in case of success the entrepreneur benefits. This situation is of course highly lucrative for the entrepreneur and highly risky for the creditor. Therefore even though the entrepreneur might have a good case the creditor almost certainly will not lend large amounts of money when the risk is divided this unequally. This example is just an illustration of the consequences that excessive debt levels might have. Although in reality such a situation is hardly observed, it might give rise to the idea that if firms face excessive debt levels and are able to find financing they also might acquire toher firms.

Graham and Harvey (2002) report that 81% of the firms have target debt ratios in order to maximize firm value. Therefore it seems unlikely that firms with excessive debt levels as described in the former paragraph are able to acquire. However, Uysal (2011) and Vermaelen and Xu (2014), do document that firms sometimes deviate from

6 their target debt ratios. Firms do this to be able to acquire other firms or to find optimal firm value after acquisitions.

Thus, firms should pursue an optimal level of debt where they obtain the benefits of having a tax shield, and therefore increase firm value, and discipline management because managers are not able to invest in firm value destroying projects. Additionally, by trading off these issues the organization becomes more efficient. On the other hand, firms do not want to become in a situation of debt overhang not being able to invest at all. In this situation the firm is forgoing positive NPV projects since it is not able to finance the project. Whereas in a situation of low leverage firms have the opportunity to invest in good projects, because of the absence of debt overhang as well as bad projects because of the increase in agency costs. Therefore, the likelihood of undertaking acquisitions by firms is higher when firms have low debt levels. Therefore I set my hypothesis accordingly.

Hypothesis 1: Firms that engage in acquisitions have lower leverage levels compared to firms that do not engage in acquisitions.

2.2 leverage and Acquisitions

There are several arguments why firms engage in acquisitions. One of the most important motives for acquisitions is synergy gain (Healy, Palepu and Ruback, 1992).

Furthermore, firms might acquire other firms for diversification reasons. In this way firms are able to reduce risk. For instance, firms are only able to acquire if they can pay for the acquisition. In order to pay for the acquisition they either have enough cash or need to borrow the money. In other words, their capital structure should be sufficient in order to be able to engage in acquisitions. As we have established the importance and the difficulty of finding the optimal capital structure in the former paragraphs it is logical that firms set a particular target capital structure and try to achieve this level.

However it is often not easy for firms to achieve this target capital structure. Moreover Uysal (2011) describes in his paper that pursuing the target capital structure not always have the priority of firms. The results of his study show that firms sometimes deviate from their capital structure in order to plan and structure acquisitions. His study finds that especially overleveraged firms relative to their target debt levels are less likely to engage in acquisitions, have smaller targets and pay lower premiums.

7 Morellec and Zhdanov (2008) investigate the role of leverage among bidding firms. The results of their study indicate that the bidder with the lowest leverage wins the acquisition contest. Furthermore they predict that the leverage of the winning bidder is below the industry average and that acquirers should lever up after takeover announcement.

Prior studies on leverage and investment are ambiguous. Whited (1992) finds that investment is dependent upon cash flows in firms with high leverage compared to firms with low leverage. Ghosh and Jain (2000) investigate whether financial leverage changes around mergers. They find weak evidence that an increase in debt capacity following from the merger is due to unused debt capacity from pre-merger years. This implies that firms prior to merger might have had lower debt levels.

Previous studies also addressed the effect of debt ratings on the likelihood of acquisitions (Harford and Uysal, 2014), since overleveraged firms are less likely to undertake acquisitions. Firms that have high leverage levels face bankruptcy costs and therefore increase the overall risk of default of the firm. It seems plausible that firms having a high debt rating (f.i. triple A) are better able to acquire firms. The authors indeed find that the likelihood of undertaking an acquisition is higher for firms with debt ratings than for firms without debt ratings. The authors relate their findings to the free cash flow hypothesis. This theory states that managers are financially constrained as a consequence of the firms’ debt market access and this prevents them from overinvesting. Due to the shortage of funds available, managers have to select which NPV projects to undertake. Because of the financial constraints firms are only capable of undertaking the highest NPV projects.

Although numerous research has investigated the optimal capital structure on the firms’ returns and the influence of leverage on firm value, still little research exists regarding the influence of leverage in takeover decisions. The existing literature is more extensive on the relation between leverage levels and acquisitions from a target perspective. For instance, a well-known and widely cited study by Palepu (1986)¹ investigated the probability of takeover targets using several firm characteristics. Not many studies investigate the relation between leverage and acquisition from the perspective of acquirers. For instance, Trahan and Shawky (1992) found that acquiring

1 Next to Palepu (1986) numerous studies focused on the characteristics of targets in acquisitions, among those studies are: Hasbrouck (1985), Walter (1994) Morellic (2004) Powell (2007).

8 firms could be identified with financial ratios. In their study they address two specific issues. First, the authors determine the characteristics of firms that have acquired other firms and second, they examine industry specific effects of acquisition announcements.

By using a US sample in the period between 1984-1986 the authors account for the merger wave that was observed a few years prior to their research.

Furthermore a study by Sorensen (2000) shows that acquiring firms are more profitable than non-acquiring firms. This is interesting for this study since it indicates that profitability is an indication for takeovers. Moreover, Sorensen (2000) relates this finding to the extent that mergers around 1996 are primarily motivated by firms that seek profit improvement, through rapid expansion of sales. Clayton and Ravid (2002) investigate the effect of leverage on bidding behavior. By making use of auction theory the authors find that as debt levels increase firms tend to reduce their bids. The lower bids give the competition incentives to reduce their bids as well. These findings indicate that debt levels have an influence on the bidding behavior of acquirers and therefore it is likely that managers set debt levels prior to acquisition. On the other hand, the findings might also indicate that firms with particular debt levels are not able to win the auction and therefore do not undertake acquisitions.

The importance of financial leverage during this period is also confirmed in other studies focused on the importance of the method of payment used in mergers. Trifts (1991), in particular, noted that financial leverage usually increased after a cash purchase and decreased after a stock acquisition. Increased leverage was associated with higher abnormal performance after the merger. Furthermore the author shows that both the degree of leverage of the bidding firm and the debt-equity ratio of the competition are important factors in the bid a firm is willing to submit. In particular, as a firm increases its leverage, the highest bid it is willing to submit declines. Next to the bidding behaviour of firms the payment methods of acquiring firms are also addressed in the literature. In a more recent study the choice of payment method is influenced by the acquirers leverage, state ownership and the amount of cash holdings in the firm (Boateng, 2014). Furthermore, Moeller, Schlingemann and Stulz (2004) find no evidence for the leverage ratio to influence the announcement returns associated with acquisition announcements. Bharadwaj and Shivdasani (2003) also do not find a relation between the amount of debt in a company’s capital structure and announcement returns. However, in some earlier work, Smith and Kim (1994) and

9 Maloney et al. (1993) provide evidence for a positive relation between the leverage ratio and abnormal returns following acquisition announcements.

2.3 Cross-border Acquisitions

In the last decade the volume of cross-border acquisitions has been growing

substantially. Erel, Liao & Weisbach (2012) document that the volume of cross-border acquisitions in 2007 was 45% of total merger volume worldwide compared to 23% in 1998. The authors furthermore highlight that the initial reasons of engaging in cross- border acquisitions are not that different from domestic acquisitions. The motives for acquisitions are discussed in the paragraph above, however cross-border acquisitions are different for several reasons. Firstly, firms might face cultural or geographical

differences compared to domestic acquisitions. Furthermore, governance systems might differ between countries. Another reason why cross-border acquisitions are different than domestic acquisition can be related to the imperfect integration of capital markets between countries. Therefore valuation differences might be a driver for cross-border acquisitions (Erel, et al, 2012). Furthermore, a reason for engaging in cross-border acquisitions is differences in taxes between countries. A firm might want to acquire a foreign firm in a country where corporate taxes are lower. In this way it is able to benefit from the tax structure in the foreign country. Furthermore, if firms want to enter a foreign market the easiest way is to acquire a firm that is already located in the foreign country (Buckley and Ghauri, 2002).

In line with the increase in cross-border acquisitions is the growth in the volume of literature related to cross-border acquisitions. For instance, Moeller and Schlingemann (2005) investigate, by using a US sample of 4430 acquisitions between 1985 and 1995 If stock announcement returns differ between domestic and cross-border acquisitions.

The theory of cross-border acquisitions I develop here, builds upon the aforementioned theories regarding leverage and domestic acquisitions. Since we have established that over leveraged firms must forgo NPV projects in their home country. It is very likely that this also accounts for NPV projects outside their country. The intuition here is why should firms go abroad and face (even) more asymmetric information.

Because when firms are not even able to undertake all positive NPV projects in the home country it is not likely that these firms will go abroad. Therefore it is reasonable to think

10 that firms acquiring cross-border have lower leverage than firms that acquire in their home country.

As the higher degree of leverage signals a higher probability of default firms might want to acquire foreign assets in order to have better access to capital markets and therefore lower risk. Bris, Brisley and Cabolis (2008) state that if target firms are acquired by foreign firms coming from countries with better shareholder protection and better accounting standards, the performance of the combined firm is likely to improve.

If we review this the other way around, the study indicates that firms from better investor protected countries benefit from acquiring outside their country.

Although the existing literature is inconsistent in revealing the effects of leverage on takeover likelihood various studies do find significant results. Lewellen (1971) states in his study that low leverage traditionally signals unused debt capacity, indicating that firm value is not maximized yet and potential benefits are therefore present.

In line with the latter, DeAngelo and DeAngelo (2007) argue that the value of financial flexibility provided by low leverage is under-appreciated in typical explorations of the trade-off model. The results of their study show that under levered equity acquirers remain under levered and that their debt levels trend upward only slowly. In contrast, cash acquirers become over levered and steadily reduce their leverage back down. The coefficient of leverage in their study is negative and significant.

Furthermore,Myers-Majluf (1984) state that value is created in mergers when firms with low financial leverage acquire firms with high financial leverage. Bruner (1988) and Uysal (2011) find evidence that bidders are relatively less leveraged. The former even finds evidence that acquirers increase their leverage levels one year after the acquisition. These findings are consistent with financing policies that state that growing firms choose lower leverage levels to be able to invest, whereas value firms choose higher leverage ratios and do not invest (Rajan and Zingales, 1995).

2.4 Leverage and cross-border acquisitions

Little prior studies relate to leverage and cross-border acquisitions. A study by Kang et al. (2000) highlighted the differences in leverage of United States and Japan acquirers.

The author describes that traditionally speaking the Japanese firms rely more on debt compared to the United States.

11 Another study related to leverage and acquisition is the study of Hennart and Park (1993) their study examines the determinants of entering foreign markets but not through engaging in acquisitions but setting up new ventures. In their study they use a leverage ratio but found no significant results.

After having established hypothesis 1 and the reasons thereof. I here develop the reasoning behind hypothesis. Since there is little research regarding the degree of leverage within firms and their likelihood of engaging in cross-border acquisitions I build further upon the theories that were the basis of hypothesis 1. Together with the literature stated in the former paragraph on cross-border acquisitions I construct hypothesis 2. If a firm faces a situation of too high leverage levels and must forgo positive NPV projects in its own country it is very unlikely that this firm will engage in cross-border acquisitions. Furthermore, since engaging in cross-border acquisitions already leads to more asymmetric information is likely that firms undertaking cross- border acquisitions have even lower leverage than firms that engage in domestic acquisitions. The reasoning behind this is that if a firms decide to undertake a cross- border acquisition this already increases the risk for the firm, because of the increase in asymmetric information. Therefore a firm does not want to have even more risk as a result of high debt levels. When taking hypothesis 1 into consideration it is likely that firms that acquire cross-border have even lower debt levels compared to firms that engage in domestic acquisitions. Therefore I state hypothesis 2 accordingly:

Hypothesis 2: Firms that engage in cross-border acquisitions have lower leverage than firms that undertake domestic acquisitions.

3. Methodology

In this section I describe the methodology used. In order to test the hypotheses and to determine whether the degree of leverage is indicative for acquisitions the tests require a probit model, of which I will elaborate on in the next paragraph.

3.1 Dependent variable

12 In this study I make use of a probit regression model. In this way I am able to determine whether the amount of leverage in a firm is a proper indication for acquisitions. I follow Espahbodi and Espahbodi (2003) that indicate that the probit regression method is proper method to use for this kind of study. I test several methods and compared their ability to predict takeovers. The probit regression model is a model where the dependent variable to be used is a dummy variable that equals one if an acquisition is made and equals zero otherwise. Throughout the study this dependent variable is used in three different ways. Firstly, I determine a model where the dependent variable equals one if an acquisition is made. I regress the equation over the entire sample.

Secondly, the dependent variable is used in order to determine the influence of leverage on takeovers for domestic acquisitions. Again, the entire sample is used however the dependent variable equals one if the acquisition is made in the United Kingdom and zero otherwise. Lastly, I estimate a third equation where the dependent variable equals one if a foreign acquisition is made and zero otherwise.

3.2 Leverage variables

The variable leverage can be interpreted in various ways and therefore it might be that several accounting measures are of influence on the degree of leverage. First, following the majority of the previous studies and the most used measure of leverage I include the debt-to-equity ratio in this study. This ratio is also used by Bruner (1988) and Trahan and Shawky (1992). The leverage measure used in the study is lagged by one year in order to allow for an impact period. Hereby I follow Uysal (2011). I make use of one key leverage measure and will use another measure to check for robustness. The key

leverage measure to be used in this study is the debt-equity ratio.

(1)

Here, LEVit refers to the degree of leverage in firm i at time t as a consequence of the total debt divided by the common equity of firm i at time t. I will check this leverage measure for robustness with another well-known leverage measure, following Jensen and Meckling (1976).

(2)

13 3.3 Control variables

Following the study of Uysal (2011) and Vermaelen and Xu (2014) several control variables are included in the regression model. In prior studies the importance of profitability is addressed by several papers, Therefore, I include the profitability as a control variable following Uysal (2011).

(3)

Here, the dependent variable PROFit refers to the profitability of the firm. This variable is measured by dividing the firms’ EBITDA² by the Total Assets. I include the profitability in the model lagged by one year in order to allow for an impact period.

Furthermore as described in the theory section, debt overhang

may prevent firms

from investing in positive NPV projects (Myers, 1977). Therefore, I include following Uysal (2011) and Erel et al. (2012), growth opportunities in the regression model. I use this control variable to account for effects of potential investment opportunities and misvaluation. A study by Bruner (1988) documents that cash balances are useful in evaluating the capital structure of firms. Furthermore the study states that an increase in debt capacity is offset by a decrease in cash balances. Moreover, Harford (1999) links this finding to acquisitions by addressing the importance of the source of financing in acquisitions. His study investigates the relation between corporate cash reserves and bidder announcement returns and finds that companies with excessive cash reserves are morelikely to attempt acquisitions. I measure this by dividing the amount of cash by total assets (CASHit) (Harford, 1999). Lastly, I include a manufacturing dummy in the sample since the majority of the acquisitions is made by manufacturing firms. The dummy variable equals one if the acquirer has 2 or 3 as first SIC code and otherwise zero.

2 EBITDA refers to the common financial accounting item that stands for, earnings before interest taxes depreciation and amortization.

14 3.4 Model specification

In order to test hypothesis 1 I use probit regression formulas as stated below. The dependent variable takes the value one if firm i undertakes an acquisition in year t and zero otherwise. I use control variables as discussed in the former paragraph and account for year effects (γ). The ε is the error term. The set-up of the models is then as follows.

In order to estimate the impact of leverage on acquisition I include the whole sample where the dependent variable is one in a year of acquisition and zero in a year where no acquisition is made. The probit regression set-up is as follows when I account for leverage-lagged variables.

(4)

In order to test the leverage effects in a domestic setting I require acquisitions to be domestic. Using these criteria the dependent variable turns one if a domestic acquisition is made and zero in a year if no acquisition is made. I make use of the same control variables as presented in equation 4. Here, I also use the one period lagged leverage variable. The control variables are also lagged one year.

(5)

In order to be able to test the second hypothesis I use the probit regression formula as stated below. Here, the dependent variable takes the value one if an acquirer undertakes a foreign acquisition and zero otherwise.

(6)

Again, this model is estimated using control variables and year fixed effects. All variables are lagged one period.

15

4. Data

Here, I describe the main dataset employed in this study and the construction thereof. I first elaborate on the reasons of choosing the United Kingdom as input for my research.

Next, I discuss the characteristics and the construction of the final sample used in this study. Afterwards, I will present an overview of the variables used and show descriptive statistics.

4.1 Sample selection

Since the greater part of the empirical literature about capital structure and acquisitions focuses on the United States it is interesting to see whether these findings also hold in a European setting. I choose the United Kingdom for several reasons. Firstly, in the United Kingdom, mergers and acquisitions are more common than in continental Europe (Hillier et al, 2012). Furthermore, the UK accounts for the large majority of European deals, as 65.3% of their 13 European country mergers are UK bidders (Faccio and Masulis, 2005). Besides, the majority of deals in the UK are cash financed in contrast to the US where most deals are stock financed (Andrade, 2001). Secondly, Rajan and Zingales (1995) found interesting results on capital structure in the G7 countries³. The authors found that firms in the United Kingdom and Germany have (on average) lower leverage. As a consequence of the better availability of data for the United Kingdom compared to Germany I choose the United Kingdom for this study.

The first major dataset employed in this study is constructed from the mergers and acquisitions database of Bureau van Dijk’s Zephyr. This database contains very detailed and extensive information regarding mergers and acquisitions. The predominant reasons for using this dataset is that it contains a long history of completed mergers and acquisitions in the United Kingdom. Furthermore the database provides information on deal year, payment methods and deal value. In order to be able to compare leverage levels of firms and obtain accounting data I make use of a second database namely, Worldscope/Datastream. The reason for using this dataset is twofold.

3 The G7 countries consist of the United States, Canada, Germany, France, United Kingdom, Japan, and Italy.

16 Firstly, this database contains information on balance sheet items, which are used to construct leverage measures in this study. Furthermore I retrieve financial data as, cash, market value, total assets and net sales from it. The second reason for using this database is that it contains financial data of all listed UK firms. This way I am able to construct a dataset that contains a benchmark of UK listed firms. For comparison between the firms that have engaged in acquisitions and firms that have not engaged in acquisitions the dataset is structured as panel data and consists of yearly observations over the period 2000-2015. I use yearly data to be able to match this data with the data obtained from Zephyr.

Since, Bureau Van Dijk’s Zephyr consists of millions of deals I first construct a raw dataset in Zephyr. Initially, Zephyr comprises 263,200 deals of firms that are listed/unlisted/delisted. For the purpose of comparability and as consequence of the availability of financial data I require the acquirers to be listed firms. I use listed acquirers since the data availability in Worldscope is on listed firms. The deal type is acquisition this reduces the sample to 139,305. When taking the sample period 2000- 2015 into account, 131,434 acquisitions remain. Only 13,077 acquisitions remain when selecting the United Kingdom as country to be studied. Lastly I account for deal status. I remove acquisitions that might not be completed due to the unavailability of this data.

The deal status of acquisitions can take the following forms; completed, completed confirmed and completed assumed after dropping the incompleted acquisitions I end up with a raw dataset consisting of 11,172 acquisitions.

After constructing the raw acquisition dataset, using the Zephyr database, I construct an acquisition dataset where the acquisitions must comprise certain criteria in order to become part of the study. Here, I elaborate on the conditions on which a firm must comply in order to become part of the acquisition dataset. Following previous studies (e.g. Hovakimian, Opler and Timan, 2001; Uysal, 2014; Vermaelen, 2014) I exclude financial firms (6000-6999) and regulated utilities (4900-4999) using Standard Industrial Classification codes (SIC). I also drop firms with sales under $ 10 million.

Furthermore, following Vermaelen (2014) I only include acquisitions of which at least 50% of the target is acquired by the bidder. This is done in order to ensure that the acquirer consolidates the target’s firm balance sheet after the transaction (Vermaelen, 2014). I also follow Moeller, Schlingemann and Stulz (2004) I drop acquisitions of which the ratio of the transaction value to total assets (acquirer) is less than 1%.

17 After eliminating firms that do not meet the restrictions I end up with a sample of 2197 acquisitions. This sample size is comparable to other studies as Uysal (2011) and Vermaelen (2014).

After merging the acquisition data obtained via Zephyr with the data retrieved from Worldscope I construct a final dataset consisting of 3900 firms and 66,300 firm years over the sample period for all UK listed firms. The sample consists of 754 firms that engaged in acquisitions during the 2000-2015 period. The other 3146 firms are used as benchmark sample. Criteria for the benchmark sample are firms that have been listed since 1999 in the UK in order to obtain financial data if those firms acquired in the year 2000. Furthermore data availability is a restriction to become part of the sample.

After drop outs due to data availability I finally end up with a sample of 1979 acquisition year observations including.

Table 1: descriptive statistics

This table present the decsriptive statistics of the variables used.

The variables are stated in their ratios as used in the regression models.

(1) (2) (3) (4) (5) (6)

Variables N mean sd min median max

Acquisitions 66219 0.023 0.166 0.000 0.000 1.00 Leverage 38233 0.556 1.847 -7.391 0.243 11.314 Profitability 36670 -0.015 0.418 -2.657 0.090 0.445 Growth 53121 2.522 5.545 -12.99 1.400 38.070 Size 34960 12.637 3.851 1.134 1.134 25.814 Cash holdings 35811 0.241 0.326 0 0.118 1.991

The list of criteria in order to become part of the sample is presented in Appendix A. Following Uysal (2011) and Vermaelen and Xu (2014) I reduce noise by winsorizing all variables at top and bottom 1 % levels. In order to capture correlation effects between the error terms I capture heteroskedasticity in all models using Huber-white method.

In the table 7 in Appendix D present a summary of the deal statistics during the sample period. After dropouts 1800 deals remain in the sample. In table 5 there is a pattern observable where the number of deals increased until the start of the financial crisis in 2007. After the crisis the amount of deals started to increase again. Furthermore a remarkable insight in the amount of cross-border acquisitions is that despite a slight

18 decrease during crisis years the amount of cross-border deals sharply increased in the last years. Slightly more than 66% of the acquisitions are financed with cash and 21 % is financed with shares and slightly over 10% is financed with debt.

5. Results

In this section the results of the (probit) regression models are discussed. First, in the following paragraph I discuss the results using the debt-equity ratio as proxy for leverage on the determination of acquisitions over the entire sample. Second, I discuss the same leverage variable for acquirers. Afterwards I present and discuss the results of several robustness checks.

5.1.1 Interpretation of regression results (hypothesis 1)

The probit regression results of estimating equation 4 are provided in table 2. Column 1 shows the results of equation 4. The motivation for using the probit model is that via this model I can observe results that distinguish between years in which firms did not undertake acquisitions and years in which firms did undertake acquisitions. To estimate equation 4 I include all firms. Thus, the 754 firms that undertake acquisitions as well as all 3146 benchmark firms. Those firms show up as either one or zero in the dependent variable. One if the firm made an acquisition in a particular year and zero otherwise.

TABLE 2: modelling likelihood of undertaking an acquisition

The table presents the results coefficients and stard errors (in parentheses) of three probit regression acquisition models. In the first column a probit model is estimated over the entire sample. The dependent variable equals one in a year if an acquisition is made. The second column presents the probit regression results for domestic acquisitions. The dependent variable equals one if a domestic acquisition is made an zero otherwise. All models include year dummies ***,**,* stand for statistical significance at the 1%,5%,10% respectively

ALL DOM FOR

Variables 1 2 3

19 Leverage

-0.027***

(0.006) -0.021***

(0.007) -0.029***

(0.008) Profitability

0.549***

(0.092) 0.426***

(0.092) 0.555***

(0.160) Growth

0.005**

(0.003) 0.001

(0.003) 0.011***

(0.003)

Cash

-0.000***

(0,000) -0.000***

(0,000) -0,000***

(0,000) Size

0.029***

(0.007)

0.037***

(0.008)

0.045***

(0.008) Manufacturing

dummy -0.078***

(0.027) -0.232***

(0.034) 0.136***

(0.035)

Year fixed effects YES YES YES

Observations 27372 26702 26379

R squared 0.053 0.070 0.023

As can been seen from table 3 in column 1, significant results were measured across the acquisition sample. The link between leverage and acquiring firms is negative and significant at the 1% level. This indicates that firms with a higher degree of leverage are less likely to engage in acquisitions, whereas firms having a low leverage are more likely to undertake acquisitions. Although the magnitude of the decrease in acquisition probability followed by an increase in leverage is little (-1,1%) the obtained result is consistent with hypothesis 1. The obtained result is in line with the study of Uysal (2011) who also reports that over levered firms are less likely to engage in acquisitions.

The coefficients relating to profitability (PROF) are positive and statistically significant at the 1% level. This result indicates that firms obtaining positive revenues are more likely to undertake acquisitions. The growth effect, measured by the market- to-book ratio (MTB) is positive in sign but statistically insignificant and therefore it does not improve the prediction power of the model. Therefore I cannot state that there is growth effect present in this model. The obtained result for the growth effect is

20 consistent with the study of Vermaelen and Xu (2014). However it is not in line with the study of Uysal (2011) who documents a slight positive relation between the market to book ratio and the likelihood of acquiring. A more surprising result is the negative but significant coefficient of the control variable cash (CASH). I estimate this by dividing the amount of cash by total assets. This regression result indicates that firms facing an increase in their amount of cash are less likely to undertake acquisitions. This result is not in line with the study of Harford and Uysal (2014) and Vermaelen and Xu (2014).

However this result is consistent with the study of Gao (2016). His study reports that firms having excess cash levels will spend more of their cash on debt reductions than on acquisitions. Therefore obtaining a coefficient with a negative sign in this model could make sense. However when taking prior studies into account I have to be prudent here.

Furthermore, the regression results of equation 4 indicate the presence of a size effect.

The size effect (SIZE) is measured by taking the natural logarithm of the net sales. The coefficient of this control variable is positive and statistically significant at the 1% level and is consistent with the study of Harford and Uysal (2014). I also account for industry effects by separating the manufacturing firms of the other firms. The dummy variable (MANU) turns one for firms of the manufacturing industry and zero otherwise. The coefficient of the manufacturing dummy is negative and significant, indicating that manufacturing firms are less likely to engage in acquisitions compared to other firms. Of all the control variables the profitability variable (PROF) has the largest explanatory power. This implies that firms facing positive revenues are more likely to undertake acquisitions.

Overall, the explanatory power of the model is denoted by the Pseudo-R² is slightly less compared to Vermaelen and Xu (2014) their study reports pseudo R- squared 7,6 and 7,9 % and Uysal (2011) documents a pseudo R-squared of 5,4%.

The results regarding the link between the degree of leverage and domestic acquisitions results are presented in the second column of table 2. Here, I use the domestic sample as dependent variable. Automatically this means that the foreign acquisitions are left out of the sample and are treated as zero. This means that the dependent variable equals one if the acquirer undertakes a domestic acquisition and equals zero otherwise. Again, I use the entire sample as input for this regression analysis in order to refrain from selection bias. This indicates that firms that never engaged in acquisitions are also part of the sample. Furthermore the dependent dummy variable

21 equals one if a domestic acquisitions is made, automatically means that the foreign acquisitions dummy is zero. As can be seen from table 3 statistically negative results were measured on the leverage indicators. Therefore the regression results indicate that leverage negatively affects the likelihood of undertaking acquisitions, which is consistent with the studies of Uysal (2011) and Vermaelen and Xu (2014). Here, the magnitude of the leverage coefficient is smaller compared to the whole sample. This result is also consistent with hypothesis 1.

The profitability coefficient of equation 5 is positive and significant at the 1% level.

Indicating that positive revenues result in a higher likelihood of undertaking acquisitions. This result is line with the studies of Uysal (2011) and Harford and Uysal (2014). Moreover, the magnitude of the profitability variable in this model is even larger compared to those studies. Again the market-to-book ratio is positive but highly insignificant. Therefore I cannot make inferences about the growth effect. The cash holding effect of equation 5 is negative as was in the model when using the entire sample. The sign of the cash holdings coefficient is even more negative compared to the model where I include all acquisitions. However the magnitude of the coefficient is very small, and therefore of little impact on the acquisition likelihood. Lastly, the size effect (SIZE) is also present in this model. The coefficient is positive and significant at the 1%

level indicating that larger firms are more likely to undertake acquisitions. The magnitude of the size coefficient is larger in equation 5 compared to the model of all acquisitions. This implies that the size effect is larger in the home country. Again, the profitability variable provides the largest explanatory power.

5.2 Interpretation of results for domestic and foreign acquisitions (hypothesis 2)

In this paragraph I discuss the results for foreign acquisitions and try to indicate whether domestic and foreign acquisitions are made by firms with different leverage levels. The results of equation 6 are provided in the third column of table 2.

The leverage coefficient is negative and significant in sign at the 1% level. The magnitude of the coefficient is larger when comparing this result to the domestic sample. This implies that firms acquiring cross-border have lower leverage levels than firms that engage in domestic acquisitions.

22 The profitability variable (PROF) is positive and the magnitude of this coefficient is larger compared to model consisting of domestic acquisitions. This implies that firms engaging in cross-border acquisitions are more profitable than domestic acquirers. In other word profitable firms are more likely to undertake cross-border acquisitions. In contrast with the former two models the market0to-book ratio here is significant at the 1% level. This result indicates that cross border acquisitions can partially be explained by firms facing sufficient market values over their book values. However the magnitude of the coefficient is very small. Again, the cash coefficient is negative indicating that cash holding firms are less likely to undertake cross-border acquisitions. The size effect is also observed in equation 6 and is positive in sign. In contrast to equations 4 and 5 the manufacturing dummy is positive in sign and significant at the 1% level this result implies that manufacturing firms are more likely to engage in cross-border acquisitions.

It is difficult to determine the statistical significance of the difference between the leverage variable of foreign acquisitions and domestic acquisitions. In order to test whether both models statistically differ from each other interaction terms should be used, however in probit regression analysis the use of interaction terms give rise to certain problems. The probit regressions do not easily deal with capturing interaction terms in the regression and therefore I must be very careful when interpreting results that compare the foreign acquisition model with the domestic acquisition model. When looking at the standard errors and the coefficients of the model variables I can determine, by multiplying the standard errors by 1.96 and adding and subtracting this from the coefficient if the variables are statistically significant from each other.

However, I must be very prudent when interpreting these results. Taking these values into account and when looking at the coefficients of the models derived with equation 5 and . Where the magnitude of the negative sign is slightly more negative for the foreign acquisition model compared to the domestic model. This implies it is more likely for firms to engage in cross-border acquisitions when their leverage is lower compared to the firms that undertake domestic acquisitions. This implied result is consistent with hypothesis 2, however I must stress that this is slight evidence and I must be very careful when comparing column 2 and 3 of table 3.

5.3 Robustness checks

23 Here, I discuss the regression results of the equations 3,4,5 by making use of a different leverage measure. I replace the leverage measure (eq. 1) which I used in the equations 4,5 and 6 by the equation 2 . Here, I test whether the results obtained in table 3 are robust to other measures of leverage. In table 4 I present the results of equations 4,5 and 6 using the alternative leverage measure.

TABLE 3: modelling likelihood of undertaking an acquisition

The table presents the coefficients and standard errors in parentheses of three probit regression acquisition models. In the first column a probit model is estimated over the entire sample. The dependent variable equals one in a year if an acquisition is made. The second column presents the probit regression results for domestic acquisitions. The dependent variable equals one if a domestic acquisition is made an zero otherwise. All models include year dummies ***,**,* stand for statistical significance at the 1%,5%,10%

respectively.

ALL DOM FOR

Variables 1 2 3

Leverage

-0.112***

(0.021) -0.117***

(0.021) -0.094***

(0.033)

Profitability 0.510***

(0.090) 0.389***

(0.091) 0.525***

(0.157)

Growth -0.006**

(0.003) -0.009**

(0.004) 0.001 (0.003) Cash

-0.000***

(0.000)

-0.000***

(0.000)

-0.000***

(0.000)

Size 0.035***

(0.007) 0.044***

(0.009) 0.048***

(0.009) Manu

-0.070**

(0.027) -0.223***

(0.035) 0.144***

(0.036)

Year fixed effects YES YES YES

Observations 26098 25428 25104

R squared 0.061 0.078 0.059